Drivers of variation in the community response to temperature change across realms
(using mixed effects models)
Collaborators: Shane Blowes, Jon Chase, Helmut Hillebrand, Michael Burrows, Amanda Bates, Uli Brose, Benoit Gauzens, Laura Antao Assistance: Katherine Lew, Josef Hauser
Introduction
- Climate change is driving a widespread reorganization of ecological communities around the world (Parmsesan & Yohe 2003, Poloczanska et al. 2013),
- but the impacts of climate change vary substantially from one location to another and among taxa (Molinos et al. 2016 NCC, Antao et al. 2020 NEE).
- Community reorganization is substantially more common than an aggregate loss or gain of species (Dornelas et al. 2014 Science, Blowes et al. 2019 Science, Hillebrand et al. 2017 J Appl Ecol)
- There are many hypotheses for why some communities are more sensitive to warming than others, including differences in
- metabolic rates (Dillon et al. 2010 Nature),
- thermal physiology (Deutsch et al. 2008 PNAS, Pinsky et al. 2019 Nature),
- microclimate availability (Burrows et al. 2019 NCC, Suggitt et al. 2018 NCC),
- species mobility (Poloczanska et al. 2013 NCC, Burrows et al. 2011 Science, Sunday et al. 2012 NCC)
- or generation time (Beaugrand et al. 2009 DSR II, Poloczanska et al. 2013 NCC),
- consumers vs. producers (Petchey et al. 1999 Nature)
- community composition (Stuart-Smith et al. 2015 Nature, Beaugrand et a. 2015 NCC, Trisos et al. 2020 Nature),
- ecosystem productivity (Thomas et al. 2017 GCB, Brett 1971 Am Zoo),
- exposure to human impacts (White & Kerr 2006 Ecography)
- and among realms (Antao et al. 2020 NEE).
- Scaling up from organismal effects to whole ecological communities is complex, and yet these scales are critical for ecosystem functioning and human well-being.
- There is a need for a comprehensive test to understand where warming is driving and is likely to drive the most dramatic community turnover
Methods
- BioTime dataset, gridded to 96 km2 hexagons, summarized as temporal turnover (Blowes)
- Temporal slope of Jaccard turnover compared to the first year
- Same for Jaccard total
- and Morisita-Horn turnover
- Explanatory variables considered for differences in rate of turnover:
- Temperature trend over the time-frame of each time-series (CRU TS 4.03 on land and in freshwater, ERSST v5 in the ocean)
- Seasonality as a metric of thermal sensitivity (Deutsch et al. 2008 PNAS). Standard deviation of monthly temperatures.
- Microclimates calculated from WorldClim and BioOracle (Laura Antao)
- Average temperature
- Body mass, collated from databases and literature searches
- Metabolic temperature, from average temperature if ectotherms (Dillon et al. 2010 Nature, Antao et al. 2020 Nat E&E)
- Mobility calculated from body mass and taxonomic group classifications of mobility mode (fly, run, swim, crawl, sessile). Fly/run/swim followed the allometric relationship in Hirt et al. 2017 Nat E&E. Crawl set at 0.1 km/hr, sessile set to 0 km/hr. Then calculated averaged within each assemblage.
- Generation time calculated from body mass and endotherm vs. ectotherm classifications, following McCoy & Gillooly 2008 ELE. Averaged across species within each assemblage.
- Consumer vs. producer classification by species
- Endotherm vs. ectotherm classification by species
- Species richness, calculated as the number of species in the assemblage
- Net primary productivity (NPP) from the merged land/ocean product produced by the Ocean Productivity group at Oregon State using methods from Zhao et al. 2005 and Behrenfeld & Falkowski 1997.
- Human impact calculated from Bowler et al. 2020 (also try data from Venter et al. 2016 and Halpern et al. 2008)
- Thermal bias calculated from Species Temperature Indices (Mike Burrows)
- Vegetation cover index, calculated from %tree cover and %non-tree veg cover (latter counted as 1/2), from vegetation continuous fields (Ruben Remelgado)
- Differences in temporal turnover (response variable) modeled with a linear mixed effects model (nlme package, lme() function). See below for details.
library(data.table) # for handling large datasets
library(ggplot2) # for some plotting
library(nlme) # for ME models
library(maps) # for map
library(gridExtra) # to combine ggplots together
library(grid) # to combine ggplots together
library(RColorBrewer)
options(width=500) # turn off most text wrapping
# tell RStudio to use project root directory as the root for this notebook. Needed since we are storing code in a separate directory.
knitr::opts_knit$set(root.dir = rprojroot::find_rstudio_root_file()) # Turnover and covariates assembled by turnover_vs_temperature_prep.Rmd
trends <- fread('output/turnover_w_covariates.csv.gz')
# set realm order
trends[, REALM := factor(REALM, levels = c('Freshwater', 'Marine', 'Terrestrial'), ordered = FALSE)]
# set up sign of temperature change
trends[, tsign := factor(sign(temptrend))]
# realm that combined Terrestrial and Freshwater, for interacting with human impact
trends[, REALM2 := REALM]
levels(trends$REALM2) = list(TerrFresh = "Freshwater", TerrFresh = "Terrestrial", Marine = "Marine")
# group Marine invertebrates/plants in with All
trends[, taxa_mod2 := taxa_mod]
trends[taxa_mod == 'Marine invertebrates/plants', taxa_mod2 := 'All']Log-transform some variables, then center and scale.
trends[, tempave.sc := scale(tempave)]
trends[, tempave_metab.sc := scale(tempave_metab)]
trends[, seas.sc := scale(seas)]
trends[, microclim.sc := scale(log(microclim))]
trends[, temptrend.sc := scale(temptrend, center = FALSE)]
trends[, temptrend_abs.sc := scale(abs(temptrend), center = FALSE)] # do not center, so that 0 is still 0 temperature change
trends[, mass.sc := scale(log(mass_mean_weight))]
trends[, speed.sc := scale(log(speed_mean_weight+1))]
trends[, lifespan.sc := scale(log(lifespan_mean_weight))]
trends[, consumerfrac.sc := scale(consfrac)]
trends[, endothermfrac.sc := scale(endofrac)]
trends[, nspp.sc := scale(log(Nspp))]
trends[, thermal_bias.sc := scale(thermal_bias)]
trends[, npp.sc := scale(log(npp))]
trends[, veg.sc := scale(log(veg+1))]
trends[, human_bowler.sc := scale(log(human_bowler+1)), by = REALM2] # separate scaling by realm
trends[REALM2 == 'TerrFresh', human_footprint.sc := scale(log(human_venter+1))]
trends[REALM2 == 'Marine', human_footprint.sc := scale(log(human_halpern))]Examine how many data points are available
Just turnover
cat('Overall # time-series: ', nrow(trends), '\n')Overall # time-series: 53013 cat('# studies: ', trends[, length(unique(STUDY_ID))], '\n')# studies: 332 cat('Data points: ', trends[, sum(nyrBT)], '\n')Data points: 293973 trends[, table(REALM)]REALM
Freshwater Marine Terrestrial
1025 48647 3341 trends[, table(taxa_mod)]taxa_mod
All Amphibians Benthos
1705 379 4679
Birds Fish Invertebrates
13741 28473 2996
Mammals Marine invertebrates/plants Plant
525 206 305
Reptiles
4 trends[, table(taxa_mod, REALM)] REALM
taxa_mod Freshwater Marine Terrestrial
All 0 1702 3
Amphibians 2 0 377
Benthos 0 4679 0
Birds 0 11099 2642
Fish 1006 27467 0
Invertebrates 15 2901 80
Mammals 0 478 47
Marine invertebrates/plants 0 206 0
Plant 1 115 189
Reptiles 1 0 3With all covariates (Bowler for human)
# the cases we can compare
apply(trends[, .(Jtutrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc, temptrend.sc, mass.sc, speed.sc, lifespan.sc, consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)], MARGIN = 2, FUN = function(x) sum(!is.na(x))) Jtutrend REALM tempave.sc tempave_metab.sc seas.sc
53013 53013 49916 49916 49916
microclim.sc temptrend.sc mass.sc speed.sc lifespan.sc
51834 49916 52820 52734 51540
consumerfrac.sc endothermfrac.sc nspp.sc thermal_bias.sc npp.sc
47534 53013 53013 49371 52863
veg.sc human_bowler.sc
52890 53013 i <- trends[, complete.cases(Jtutrend, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc, temptrend.sc, mass.sc, speed.sc, lifespan.sc, consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
cat('Overall # time-series: ', sum(i), '\n')Overall # time-series: 43493 cat('# studies: ', trends[i, length(unique(STUDY_ID))], '\n')# studies: 235 cat('Data points: ', trends[i, sum(nyrBT)], '\n')Data points: 221956 trends[i, table(REALM)]REALM
Freshwater Marine Terrestrial
978 39738 2777 trends[i, table(taxa_mod)]taxa_mod
All Amphibians Benthos Birds Fish Invertebrates Mammals
521 12 590 11753 27345 2559 515
Plant Reptiles
196 2 trends[i, table(taxa_mod, REALM)] REALM
taxa_mod Freshwater Marine Terrestrial
All 0 520 1
Amphibians 2 0 10
Benthos 0 590 0
Birds 0 9221 2532
Fish 966 26379 0
Invertebrates 9 2484 66
Mammals 0 477 38
Plant 1 67 128
Reptiles 0 0 2Choose the variance structure for mixed effects models
Try combinations of
- variance scaled to a power of the number of years in the community time-series
- variance scaled to a power of the abs temperature trend
- random intercept for taxa_mod
- random intercept for STUDY_ID
- random slope (abs temperature trend) for taxa_mod
- random slope (abs temperature trend) for STUDY_ID
- random intercept for rarefyID (for overdispersion)
And choose the one with lowest AIC (not run: takes a long time)
# fit models for variance structure
fixed <- formula(Jtutrend ~ REALM + tempave_metab.sc + seas.sc + microclim.sc + npp.sc + temptrend_abs.sc +
mass.sc + speed.sc + lifespan.sc + consumerfrac.sc + thermal_bias.sc)
i <- trends[, complete.cases(Jtutrend, REALM, tempave_metab.sc, seas.sc, microclim.sc, npp.sc, temptrend_abs.sc,
mass.sc, speed.sc, lifespan.sc, consumerfrac.sc, thermal_bias.sc)]
mods <- vector('list', 0)
mods[[1]] <- gls(fixed, data = trends[i,])
mods[[2]] <- gls(fixed, data = trends[i,], weights = varPower(-0.5, ~nyrBT))
mods[[3]] <- gls(fixed, data = trends[i,], weights = varPower(0.5, ~ abs(temptrend)))
mods[[4]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2, control = lmeControl(opt = "optim"))
mods[[5]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID, control = lmeControl(opt = "optim"))
mods[[6]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2/STUDY_ID, control = lmeControl(opt = "optim"))
mods[[7]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID/rarefyID, control = lmeControl(opt = "optim"))
mods[[8]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2/STUDY_ID/rarefyID, control = lmeControl(opt = "optim"))
mods[[9]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc | taxa_mod)
mods[[10]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc | STUDY_ID)
mods[[11]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc | taxa_mod2/STUDY_ID, control = lmeControl(opt = "optim"))
mods[[12]] <- lme(fixed, data = trends[i,], random = list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)) # includes overdispersion. new formula so that random slope is only for study level (not enough data to extend to rarefyID).
mods[[13]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)) # 30+ min to fit
mods[[14]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID, weights = varPower(-0.5, ~nyrBT))
mods[[15]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2, weights = varPower(-0.5, ~nyrBT))
mods[[16]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2/STUDY_ID, weights = varPower(-0.5, ~nyrBT))
mods[[17]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID/rarefyID, weights = varPower(-0.5, ~nyrBT))
mods[[18]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2/STUDY_ID/rarefyID, weights = varPower(-0.5, ~nyrBT))
mods[[19]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc|STUDY_ID, weights = varPower(-0.5, ~nyrBT))
mods[[20]] <- lme(fixed, data = trends[i,], random = list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1), weights = varPower(-0.5, ~nyrBT))
mods[[21]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ 1), weights = varPower(-0.5, ~nyrBT))
mods[[22]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ 1, rarefyID = ~1), weights = varPower(-0.5, ~nyrBT))
mods[[23]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ temptrend_abs.sc), weights = varPower(-0.5, ~nyrBT)) # singular precision warning with lmeControl(opt = 'optim') and convergence error without
mods[[24]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1), weights = varPower(-0.5, ~nyrBT)) # singular precision warning with lmeControl(opt = 'optim') and convergence error without
mods[[25]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2, weights = varPower(-0.5, ~abs(temptrend)))
mods[[26]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID, weights = varPower(-0.5, ~abs(temptrend)))
mods[[27]] <- lme(fixed, data = trends[i,], random = ~1|STUDY_ID/rarefyID, weights = varPower(-0.5, ~abs(temptrend)))
mods[[28]] <- lme(fixed, data = trends[i,], random = ~1|taxa_mod2/STUDY_ID/rarefyID, weights = varPower(-0.5, ~abs(temptrend)))
mods[[29]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc|STUDY_ID, weights = varPower(-0.5, ~abs(temptrend)))
mods[[30]] <- lme(fixed, data = trends[i,], random = ~temptrend_abs.sc|taxa_mod2/STUDY_ID, weights = varPower(-0.5, ~abs(temptrend)), control = lmeControl(opt = "optim"))
mods[[31]] <- lme(fixed, data = trends[i,], random = list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1), weights = varPower(-0.5, ~abs(temptrend)))
mods[[32]] <- lme(fixed, data = trends[i,], random = list(taxa_mod2 = ~ temptrend_abs.sc, STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1), weights = varPower(-0.5, ~abs(temptrend)), control = lmeControl(opt = "optim")) # singular precision warning
aics <- sapply(mods, AIC)
minaics <- aics - min(aics)
minaics
which.min(aics)Chooses the random slopes (temptrend_abs) & intercepts for STUDY_ID, overdispersion, and variance scaled to number of years. We haven’t dealt with potential testing on the boundary issues here yet.
Results
Where do we have data?
world <- map_data('world')
ggplot(world, aes(x = long, y = lat, group = group)) +
geom_polygon(fill = 'lightgray', color = 'white') +
geom_point(data = trends, aes(rarefyID_x, rarefyID_y, group = REALM, color = REALM), size = 0.5, alpha = 0.4) +
scale_color_brewer(palette="Set1", name = 'Realm') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.key=element_blank(),
axis.text=element_text(size=16),
axis.title=element_text(size=20)) +
labs(x = 'Longitude (°)', y = 'Latitude (°)')
Mostly northern hemisphere, but spread all over. No so much in Africa or much of Asia.
Average rates of turnover (with year 1)
trends[abs(temptrend) >= 0.5, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # turnover per year for locations changing temperaturetrends[abs(temptrend) < 0.1, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # not changing temperaturetrends[temptrend >= 0.5, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # warmingtrends[temptrend <= -0.5, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # cooling
trends[abs(temptrend) >= 0.5 & abs(rarefyID_y) < 35, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # tropics and sub-tropicstrends[abs(temptrend) >= 0.5 & abs(rarefyID_y) >= 35 & abs(rarefyID_y) < 66.56339, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # temperatetrends[abs(temptrend) >= 0.5 & abs(rarefyID_y) >= 66.56339, .(ave = mean(Jtutrend), sd = sd(Jtutrend)/sqrt(.N))] # arcticAverage rates of turnover (without year 1)
trends[abs(temptrend) >= 0.5, .(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # turnover per year for locations changing temperaturetrends[abs(temptrend) < 0.1, .(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # not changing temperaturetrends[temptrend >= 0.5, .(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # warmingtrends[temptrend <= -0.5, .(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # cooling
trends[abs(temptrend) >= 0.5 & abs(rarefyID_y) < 35,
.(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # tropics and sub-tropicstrends[abs(temptrend) >= 0.5 & abs(rarefyID_y) >= 35 & abs(rarefyID_y) < 66.56339,
.(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # temperatetrends[abs(temptrend) >= 0.5 & abs(rarefyID_y) >= 66.56339,
.(ave = mean(Jtutrendrem0, na.rm=TRUE),
sd = sd(Jtutrendrem0, na.rm=TRUE)/sqrt(.N))] # arcticTemperature-only model (Jtutrend, Jbetatrend, Horntrend)
With year 1
i <- trends[, complete.cases(Jtutrend, REALM, temptrend)]
randef <- list(STUDY_ID = ~ abs(temptrend), rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
if(file.exists('temp/modonlyTtrend.rds')){
modonlyTtrend <- readRDS('temp/modonlyTtrend.rds')
} else {
modonlyTtrend <- lme(Jtutrend ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i,], method = 'REML')
saveRDS(modonlyTtrend, file = 'temp/modonlyTtrend.rds')
}
i2 <- trends[, complete.cases(Jbetatrend, REALM, temptrend)]
if(file.exists('temp/modonlyTtrendJbeta.rds')){
modonlyTtrendJbeta <- readRDS('temp/modonlyTtrendJbeta.rds')
} else {
modonlyTtrendJbeta <- lme(Jbetatrend ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i2,], method = 'REML',
control=lmeControl(msMaxIter = 100, maxIter = 100))
saveRDS(modonlyTtrendJbeta, file = 'temp/modonlyTtrendJbeta.rds')
}
i3 <- trends[, complete.cases(Horntrend, REALM, temptrend)]
if(file.exists('temp/modonlyTtrendHorn.rds')){
modonlyTtrendHorn <- readRDS('temp/modonlyTtrendHorn.rds')
} else {
modonlyTtrendHorn <- lme(Horntrend ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i3,], method = 'REML')
saveRDS(modonlyTtrendHorn, file = 'temp/modonlyTtrendHorn.rds')
}
summary(modonlyTtrend)Linear mixed-effects model fit by REML
Data: trends[i, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.04375688 (Intr)
abs(temptrend) 0.29351811 -0.123
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.001792571 0.2893301
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.223838
Fixed effects: Jtutrend ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.424
REALMMarine -0.934 0.396
REALMTerrestrial -0.923 0.391 0.862
abs(temptrend):REALMMarine 0.407 -0.960 -0.407 -0.375
abs(temptrend):REALMTerrestrial 0.387 -0.913 -0.362 -0.433 0.876
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-8.38476416 -0.29781039 0.08738927 0.54929160 6.52538535
Number of Observations: 49916
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
317 49916 summary(modonlyTtrendJbeta)Linear mixed-effects model fit by REML
Data: trends[i2, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.05691594 (Intr)
abs(temptrend) 0.34739161 -0.16
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 1.28009e-06 0.3525846
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.394679
Fixed effects: Jbetatrend ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.428
REALMMarine -0.933 0.399
REALMTerrestrial -0.926 0.396 0.864
abs(temptrend):REALMMarine 0.411 -0.961 -0.411 -0.380
abs(temptrend):REALMTerrestrial 0.390 -0.913 -0.364 -0.437 0.877
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-7.3578768 -0.1587203 0.2033006 0.6705745 6.5270460
Number of Observations: 49916
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
317 49916 summary(modonlyTtrendHorn)Linear mixed-effects model fit by REML
Data: trends[i3, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.0549979 (Intr)
abs(temptrend) 0.3228314 -0.095
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.007989212 0.3034899
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.217439
Fixed effects: Horntrend ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.399
REALMMarine -0.921 0.368
REALMTerrestrial -0.919 0.367 0.847
abs(temptrend):REALMMarine 0.381 -0.954 -0.378 -0.350
abs(temptrend):REALMTerrestrial 0.363 -0.908 -0.334 -0.408 0.867
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-6.21250285 -0.32193926 0.06521414 0.53688251 6.03752738
Number of Observations: 48800
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
276 48800 Without year 1
i4 <- trends[, complete.cases(Jtutrendrem0, REALM, temptrend)]
randef <- list(STUDY_ID = ~ abs(temptrend), rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
if(file.exists('temp/modonlyTtrendrem0.rds')){
modonlyTtrendrem0 <- readRDS('temp/modonlyTtrendrem0.rds')
} else {
modonlyTtrendrem0 <- lme(Jtutrendrem0 ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i4,], method = 'REML')
saveRDS(modonlyTtrendrem0, file = 'temp/modonlyTtrendrem0.rds')
}
i5 <- trends[, complete.cases(Jbetatrendrem0, REALM, temptrend)]
if(file.exists('temp/modonlyTtrendJbetarem0.rds')){
modonlyTtrendJbetarem0 <- readRDS('temp/modonlyTtrendJbetarem0.rds')
} else {
modonlyTtrendJbetarem0 <- lme(Jbetatrendrem0 ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i5,], method = 'REML',
control=lmeControl(msMaxIter = 100, maxIter = 100))
saveRDS(modonlyTtrendJbetarem0, file = 'temp/modonlyTtrendJbetarem0.rds')
}
i6 <- trends[, complete.cases(Horntrendrem0, REALM, temptrend)]
if(file.exists('temp/modonlyTtrendHornrem0.rds')){
modonlyTtrendHornrem0 <- readRDS('temp/modonlyTtrendHornrem0.rds')
} else {
modonlyTtrendHornrem0 <- lme(Horntrendrem0 ~ abs(temptrend)*REALM,
random = randef, weights = varef, data = trends[i6,], method = 'REML')
saveRDS(modonlyTtrendHornrem0, file = 'temp/modonlyTtrendHornrem0.rds')
}
summary(modonlyTtrendrem0)Linear mixed-effects model fit by REML
Data: trends[i4, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.007657053 (Intr)
abs(temptrend) 0.197650298 -0.935
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.01097849 2.000161
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-2.121022
Fixed effects: Jtutrendrem0 ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.807
REALMMarine -0.947 0.764
REALMTerrestrial -0.901 0.727 0.853
abs(temptrend):REALMMarine 0.766 -0.949 -0.814 -0.690
abs(temptrend):REALMTerrestrial 0.733 -0.908 -0.693 -0.810 0.862
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-6.57101792 -0.22892007 -0.01904447 0.27434430 5.64177564
Number of Observations: 36747
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
292 36747 summary(modonlyTtrendJbetarem0)Linear mixed-effects model fit by REML
Data: trends[i5, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.005965236 (Intr)
abs(temptrend) 0.160013131 -0.038
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.003861919 0.9688156
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.887978
Fixed effects: Jbetatrendrem0 ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.543
REALMMarine -0.941 0.511
REALMTerrestrial -0.906 0.492 0.852
abs(temptrend):REALMMarine 0.515 -0.948 -0.524 -0.467
abs(temptrend):REALMTerrestrial 0.494 -0.908 -0.464 -0.542 0.861
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-8.57430409 -0.30896466 -0.02462766 0.32539894 8.23224466
Number of Observations: 36747
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
292 36747 summary(modonlyTtrendHornrem0)Linear mixed-effects model fit by REML
Data: trends[i6, ]
Random effects:
Formula: ~abs(temptrend) | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.01201968 (Intr)
abs(temptrend) 0.25314398 0.016
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.01882813 2.441788
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-2.318769
Fixed effects: Horntrendrem0 ~ abs(temptrend) * REALM Correlation:
(Intr) abs(t) REALMM REALMT a():REALMM
abs(temptrend) -0.472
REALMMarine -0.944 0.446
REALMTerrestrial -0.899 0.424 0.848
abs(temptrend):REALMMarine 0.447 -0.945 -0.455 -0.401
abs(temptrend):REALMTerrestrial 0.428 -0.906 -0.404 -0.473 0.856
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-5.75457482 -0.22483445 -0.02254226 0.23615730 5.73488343
Number of Observations: 36005
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
257 36005 Plot the temp-only coefficients
colors <- brewer.pal(6, 'Dark2')
# make table of coefficients
coefs <- as.data.frame(summary(modonlyTtrend)$tTable)
coefs2 <- as.data.frame(summary(modonlyTtrendJbeta)$tTable)
coefs3 <- as.data.frame(summary(modonlyTtrendHorn)$tTable)
coefs4 <- as.data.frame(summary(modonlyTtrendrem0)$tTable)
coefs5 <- as.data.frame(summary(modonlyTtrendJbetarem0)$tTable)
coefs6 <- as.data.frame(summary(modonlyTtrendHornrem0)$tTable)
coefs$mod <- 'Jtu'
coefs2$mod <- 'Jbeta'
coefs3$mod <- 'Horn'
coefs4$mod <- 'Jturem0'
coefs5$mod <- 'Jbetarem0'
coefs6$mod <- 'Hornrem0'
rows1 <- which(grepl('temptrend', rownames(coefs))) # extract temperature effect
cols <- c('Value', 'Std.Error', 'mod')
allcoefs <- rbind(coefs[rows1, cols], coefs2[rows1, cols], coefs3[rows1, cols],
coefs4[rows1, cols], coefs5[rows1, cols], coefs6[rows1, cols])
allcoefs$Value[grepl('REALMMarine', rownames(allcoefs))] <-
allcoefs$Value[grepl('REALMMarine', rownames(allcoefs))] +
allcoefs$Value[!grepl('REALM', rownames(allcoefs))] # add intercept to marine effects
allcoefs$Value[grepl('REALMTerrestrial', rownames(allcoefs))] <-
allcoefs$Value[grepl('REALMTerrestrial', rownames(allcoefs))] +
allcoefs$Value[!grepl('REALM', rownames(allcoefs))] # add intercept to terrestrial effects
allcoefs$lCI <- allcoefs$Value - allcoefs$Std.Error # lower confidence interval
allcoefs$uCI <- allcoefs$Value + allcoefs$Std.Error
allcoefs$y <- c(3, 2, 1) + rep(c(0, -0.1, -0.2, -0.3, -0.4, -0.5), c(3, 3, 3, 3, 3, 3)) # y-values
allcoefs$col <- c(rep(colors[1], 3), rep(colors[2], 3), rep(colors[3], 3),
rep(colors[4], 3), rep(colors[5], 3), rep(colors[6], 3))
allcoefs$realm <- rep(c('Freshwater', 'Marine', 'Terrestrial'), 6)
par(las = 1, mai = c(0.8, 2, 0.1, 0.1))
plot(0,0, col = 'white', xlim=c(-0.1, 0.85), ylim = c(0.5,3),
yaxt='n', xlab = 'Turnover per |°C/yr|', ylab ='')
axis(2, at = 3:1, labels = c('Freshwater', 'Marine', 'Terrestrial'), cex.axis = 0.7)abline(v = 0, col = 'grey')
for(i in 1:nrow(allcoefs)){
with(allcoefs[i, ], points(Value, y, pch = 16, col = col))
with(allcoefs[i, ], lines(x = c(lCI, uCI), y = c(y, y), col = col))
}legend('bottomright', col = colors, lwd = 1, pch = 16,
legend = c('Jaccard turnover', 'Jaccard total', 'Horn-Morisita',
'Jaccard turnover rem0', 'Jaccard total rem0', 'Horn-Morisita rem0'))
Nicer plots of turnover vs. temperature data and model fit
Scatterplot, violin plots, and coefficient plots all together Off for now
# on macbook: fig.width=3, fig.height=2.375, fig.retina=3, out.width=3, out.height=2.375
# on external monitor: fig.width=6, fig.height=4.5
trends[temptrend <= -0.5, temptrendtext := 'Cooling']
trends[abs(temptrend) <= 0.1, temptrendtext := 'Stable']
trends[temptrend >= 0.5, temptrendtext := 'Warming']
trends[abs(rarefyID_y) < 35, latzone := 'Subtropics']
trends[abs(rarefyID_y) >= 35 & abs(rarefyID_x) < 66.56339, latzone := 'Temperate']
trends[abs(rarefyID_y) >= 66.56339, latzone := 'Polar']
trends[, latzone := factor(latzone, levels = c('Subtropics', 'Temperate', 'Polar'))]
p1 <- ggplot(trends, aes(temptrend, Jtutrend, color = REALM, fill = REALM, size = nyrBT)) +
geom_point(na.rm = TRUE, shape = 16, alpha = 0.1) +
geom_smooth(data=subset(trends, abs(temptrend) < 0.75), method = 'gam', formula = y ~ s(x, bs = "cs"),
na.rm = TRUE) +
scale_color_brewer(palette="Set1", name = 'Realm') +
scale_fill_brewer(palette="Set1", name = 'Realm') +
labs(x = 'Temperature trend (°C/yr)', y = 'Jaccard turnover', tag = 'A') +
scale_size_continuous(range = c(1, 8), breaks = c(2, 5, 20)) +
guides(size = guide_legend(title = 'Years',
override.aes = list(linetype=0, fill = NA, alpha = 1))) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.key=element_blank(),
legend.key.size = unit(0.5,"line"),
axis.text=element_text(size=8),
axis.title=element_text(size=10))
p2 <- ggplot(trends[!is.na(temptrendtext), ], aes(temptrendtext, Jtutrend)) +
geom_violin(draw_quantiles = c(0.25, 0.5, 0.75), fill = 'grey') +
labs(x = '', y = 'Jaccard turnover', tag = 'B') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.key=element_blank(),
axis.text=element_text(size=8),
axis.title=element_text(size=10))
p3 <- ggplot(trends[abs(temptrend) >= 0.5 & !is.na(latzone), ], aes(latzone, Jtutrend)) +
geom_violin(draw_quantiles = c(0.25, 0.5, 0.75), fill = 'grey') +
labs(x = '', y = '', tag = 'C') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.key=element_blank(),
axis.text=element_text(size=7),
axis.title=element_text(size=10))
p4 <- ggplot(allcoefs, aes(Value, y, group = mod, color = mod)) +
geom_errorbarh(aes(xmin = lCI, xmax = uCI, height = 0)) +
geom_point() +
labs(x = expression(atop('Temperature change effect', '(Turnover '~degree*'C'^'-1'*')')), y = '', tag = 'D') +
scale_color_grey() +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='none',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_cartesian(xlim =c(0, 1)) +
scale_y_continuous(name = '', breaks = c(1, 2, 3), labels = c('Terrestrial', 'Marine', 'Freshwater'))
grid.arrange(p1, p2, p3, p4, ncol = 3, layout_matrix = rbind(c(1,1,1), c(2,3,4)),
heights=c(unit(0.66, "npc"), unit(0.34, "npc")))Full models
Try static covariates plus interactions of abs temperature trend with each covariate:
- realm
- speed
- mass
- lifespan
- average metabolic temperature
- consumer fraction
- endotherm fraction
- environmental temperature
- seasonality
- microclimates
- thermal bias
- NPP
- vegetation
- human footprint
Except for thermal bias: interact with temperature trend (not abs)
Fit full models
Full model for Jaccard turnover
Try Bowler or Venter/Halpern human impact
# using Bowler for human impact
i <- trends[, complete.cases(Jtutrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
randef <- list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
if(file.exists('temp/modTfull1.rds')){
modTfull1 <- readRDS('temp/modTfull1.rds')
} else {
modTfull1 <- lme(Jtutrend ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_bowler.sc:REALM2,
random = randef, weights = varef, data = trends[i,], method = 'REML')
saveRDS(modTfull1, file = 'temp/modTfull1.rds')
}
# using Venter/Halpern for human impact
i1.2 <- trends[, complete.cases(Jtutrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_footprint.sc)]
randef <- list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
if(file.exists('temp/modTfullfootprint.rds')){
modTfullfootprint <- readRDS('temp/modTfullfootprint.rds')
} else {
modTfullfootprint <- lme(Jtutrend ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_footprint.sc:REALM2,
random = randef, weights = varef, data = trends[i1.2,], method = 'REML')
saveRDS(modTfullfootprint, file = 'temp/modTfullfootprint.rds')
}
summary(modTfull1)Linear mixed-effects model fit by REML
Data: trends[i, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.04686628 (Intr)
temptrend_abs.sc 0.03743575 -0.109
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.000833882 0.2868584
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.239456
Fixed effects: Jtutrend ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.408
REALMMarine -0.928 0.383
REALMTerrestrial -0.865 0.350 0.801
tsign1 -0.037 0.022 0.009 0.012
tempave.sc -0.003 0.001 -0.002 -0.002 -0.071
tempave_metab.sc 0.052 -0.023 -0.045 -0.041 -0.023 -0.469
seas.sc -0.045 0.042 0.063 -0.014 -0.109 0.183 -0.099
microclim.sc -0.016 0.022 0.025 -0.008 0.012 -0.042 -0.026 0.142
mass.sc 0.014 0.009 -0.001 0.001 0.020 0.026 -0.498 0.000
speed.sc 0.002 -0.005 0.013 0.021 -0.047 0.103 -0.279 0.100
lifespan.sc 0.032 -0.022 -0.027 -0.044 -0.015 -0.087 0.707 0.013
consumerfrac.sc -0.017 0.009 0.030 0.272 0.017 -0.019 -0.038 -0.043
endothermfrac.sc 0.146 -0.050 -0.074 -0.308 0.008 0.142 -0.165 0.027
nspp.sc -0.009 -0.017 -0.013 -0.027 -0.039 0.004 -0.015 -0.045
npp.sc 0.010 0.006 -0.014 0.014 0.014 -0.276 0.213 -0.147
veg.sc -0.109 0.141 0.113 0.000 -0.016 0.145 -0.104 0.130
temptrend_abs.sc:REALMMarine 0.392 -0.960 -0.394 -0.330 -0.005 -0.001 0.019 -0.055
temptrend_abs.sc:REALMTerrestrial 0.340 -0.843 -0.314 -0.415 -0.005 -0.012 0.022 0.038
temptrend_abs.sc:tsign1 0.019 -0.046 -0.007 -0.005 -0.486 0.044 0.000 0.014
temptrend_abs.sc:tempave.sc 0.003 -0.009 -0.006 -0.004 0.048 -0.595 0.309 -0.054
temptrend_abs.sc:tempave_metab.sc -0.014 0.027 0.013 0.014 -0.011 0.326 -0.582 -0.010
temptrend_abs.sc:seas.sc 0.021 -0.061 -0.029 0.029 0.043 -0.010 -0.005 -0.618
temptrend_abs.sc:microclim.sc 0.009 -0.030 -0.012 0.010 -0.040 0.075 -0.012 -0.065
temptrend_abs.sc:mass.sc 0.007 -0.015 -0.009 -0.010 -0.014 0.024 0.265 0.007
temptrend_abs.sc:speed.sc -0.005 0.020 -0.004 -0.002 0.024 -0.087 0.129 -0.032
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.034
speed.sc 0.068 0.140
lifespan.sc -0.009 -0.774 -0.454
consumerfrac.sc -0.007 0.015 -0.070 -0.047
endothermfrac.sc 0.011 -0.014 -0.169 0.067 -0.331
nspp.sc -0.099 -0.214 0.007 0.145 -0.002 0.085
npp.sc -0.157 -0.041 0.047 0.033 -0.028 -0.083 -0.180
veg.sc 0.001 0.033 0.046 -0.026 -0.049 0.007 0.004 -0.283
temptrend_abs.sc:REALMMarine -0.026 -0.014 -0.007 0.017 -0.010 0.033 0.026 0.011
temptrend_abs.sc:REALMTerrestrial 0.015 -0.014 -0.007 0.027 -0.101 0.134 0.042 -0.028
temptrend_abs.sc:tsign1 -0.037 -0.013 0.017 0.007 -0.006 0.004 0.026 -0.049
temptrend_abs.sc:tempave.sc 0.078 0.024 -0.059 0.029 -0.013 -0.109 -0.016 0.153
temptrend_abs.sc:tempave_metab.sc -0.029 0.250 0.078 -0.346 0.058 0.134 -0.009 -0.139
temptrend_abs.sc:seas.sc -0.092 0.008 -0.035 -0.008 0.031 0.011 0.030 0.081
temptrend_abs.sc:microclim.sc -0.673 -0.015 -0.027 0.008 0.001 0.007 0.039 0.127
temptrend_abs.sc:mass.sc -0.010 -0.552 -0.030 0.436 -0.011 0.019 0.130 0.006
temptrend_abs.sc:speed.sc -0.036 -0.045 -0.536 0.187 0.031 0.092 -0.021 0.003
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.155
temptrend_abs.sc:REALMTerrestrial 0.009 0.796
temptrend_abs.sc:tsign1 -0.001 0.015 0.007
temptrend_abs.sc:tempave.sc -0.137 0.005 0.024 -0.069
temptrend_abs.sc:tempave_metab.sc 0.064 -0.022 -0.037 0.016 -0.441
temptrend_abs.sc:seas.sc -0.131 0.084 -0.078 -0.019 0.048
temptrend_abs.sc:microclim.sc 0.026 0.034 -0.032 0.048 -0.132
temptrend_abs.sc:mass.sc -0.004 0.021 0.027 0.024 -0.075
temptrend_abs.sc:speed.sc -0.052 0.003 0.001 -0.005 0.157
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc 0.047
temptrend_abs.sc:microclim.sc 0.108 0.136
temptrend_abs.sc:mass.sc -0.471 -0.044 -0.011
temptrend_abs.sc:speed.sc -0.223 0.069 0.041
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.059
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-8.34716400 -0.35364009 0.05228794 0.52742556 6.59480022
Number of Observations: 43493
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
235 43493 summary(modTfullfootprint)Linear mixed-effects model fit by REML
Data: trends[i1.2, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.04731816 (Intr)
temptrend_abs.sc 0.03945538 -0.122
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 2.288017e-06 0.2861346
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.238289
Fixed effects: Jtutrend ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_footprint.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_.
temptrend_abs.sc -0.417
REALMMarine -0.925 0.390
REALMTerrestrial -0.865 0.359 0.799
tsign1 -0.036 0.020 0.008 0.012
tempave.sc -0.011 0.009 0.005 -0.005 -0.063
tempave_metab.sc 0.057 -0.028 -0.049 -0.039 -0.027 -0.474
seas.sc -0.047 0.034 0.064 -0.013 -0.113 0.192 -0.109
microclim.sc -0.017 0.016 0.028 -0.007 0.004 0.021 -0.055
mass.sc 0.015 0.008 -0.003 0.001 0.021 0.037 -0.501
speed.sc 0.003 -0.006 0.013 0.020 -0.045 0.099 -0.272
lifespan.sc 0.032 -0.022 -0.026 -0.044 -0.014 -0.099 0.710
consumerfrac.sc -0.015 0.007 0.031 0.263 0.016 -0.026 -0.036
endothermfrac.sc 0.145 -0.051 -0.076 -0.312 0.010 0.134 -0.163
nspp.sc -0.008 -0.017 -0.012 -0.028 -0.038 0.018 -0.019
npp.sc 0.018 0.001 -0.024 0.013 0.015 -0.336 0.252
veg.sc -0.114 0.133 0.116 0.003 -0.018 0.195 -0.139
temptrend_abs.sc:REALMMarine 0.400 -0.958 -0.401 -0.339 -0.004 -0.011 0.024
temptrend_abs.sc:REALMTerrestrial 0.346 -0.845 -0.318 -0.420 -0.006 -0.008 0.018
temptrend_abs.sc:tsign1 0.018 -0.043 -0.005 -0.005 -0.486 0.044 0.000
temptrend_abs.sc:tempave.sc 0.008 -0.023 -0.012 -0.002 0.044 -0.604 0.312
temptrend_abs.sc:tempave_metab.sc -0.018 0.035 0.017 0.011 -0.008 0.332 -0.585
temptrend_abs.sc:seas.sc 0.018 -0.048 -0.026 0.025 0.051 -0.037 0.006
temptrend_abs.sc:microclim.sc 0.006 -0.016 -0.011 0.008 -0.039 0.053 0.000
temptrend_abs.sc:mass.sc 0.006 -0.013 -0.008 -0.010 -0.014 0.021 0.266
temptrend_abs.sc:speed.sc -0.005 0.020 -0.006 -0.002 0.025 -0.090 0.127
ses.sc mcrcl. mss.sc spd.sc lfspn. cnsmr. endth.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc 0.183
mass.sc -0.006 0.011
speed.sc 0.093 0.071 0.137
lifespan.sc 0.014 0.006 -0.774 -0.449
consumerfrac.sc -0.041 0.003 0.016 -0.067 -0.049
endothermfrac.sc 0.029 0.027 -0.010 -0.168 0.063 -0.329
nspp.sc -0.037 -0.104 -0.217 0.010 0.150 -0.003 0.090
npp.sc -0.208 -0.210 -0.042 0.048 0.031 -0.033 -0.099
veg.sc 0.126 0.018 0.036 0.042 -0.028 -0.052 0.018
temptrend_abs.sc:REALMMarine -0.046 -0.020 -0.012 -0.008 0.016 -0.009 0.032
temptrend_abs.sc:REALMTerrestrial 0.034 0.016 -0.015 -0.006 0.028 -0.095 0.139
temptrend_abs.sc:tsign1 0.022 -0.033 -0.012 0.017 0.006 -0.004 0.003
temptrend_abs.sc:tempave.sc -0.070 0.053 0.019 -0.059 0.035 -0.013 -0.106
temptrend_abs.sc:tempave_metab.sc 0.002 -0.014 0.251 0.076 -0.350 0.055 0.135
temptrend_abs.sc:seas.sc -0.617 -0.089 0.017 -0.039 -0.015 0.029 0.006
temptrend_abs.sc:microclim.sc -0.060 -0.703 -0.006 -0.034 0.001 -0.002 0.001
temptrend_abs.sc:mass.sc 0.014 0.001 -0.553 -0.026 0.436 -0.011 0.017
temptrend_abs.sc:speed.sc -0.037 -0.040 -0.041 -0.537 0.182 0.030 0.091
nspp.s npp.sc veg.sc t_.:REALMM t_.:REALMT tm_.:1
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc -0.188
veg.sc 0.010 -0.331
temptrend_abs.sc:REALMMarine 0.025 0.016 -0.145
temptrend_abs.sc:REALMTerrestrial 0.042 -0.027 0.000 0.797
temptrend_abs.sc:tsign1 0.027 -0.049 0.009 0.011 0.010
temptrend_abs.sc:tempave.sc -0.022 0.183 -0.189 0.021 0.019 -0.065
temptrend_abs.sc:tempave_metab.sc -0.005 -0.164 0.101 -0.030 -0.032 0.012
temptrend_abs.sc:seas.sc 0.030 0.107 -0.098 0.069 -0.071 -0.031
temptrend_abs.sc:microclim.sc 0.040 0.153 0.051 0.021 -0.030 0.041
temptrend_abs.sc:mass.sc 0.133 0.009 -0.002 0.018 0.028 0.023
temptrend_abs.sc:speed.sc -0.024 0.002 -0.052 0.004 0.000 -0.005
tmptrnd_bs.sc:t. tm_.:_. tmptrnd_bs.sc:ss.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc -0.435
temptrend_abs.sc:seas.sc 0.069 0.040
temptrend_abs.sc:microclim.sc -0.106 0.099 0.110
temptrend_abs.sc:mass.sc -0.070 -0.473 -0.056
temptrend_abs.sc:speed.sc 0.160 -0.220 0.075
tmptrnd_bs.sc:mc. tmptrnd_bs.sc:ms.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc -0.024
temptrend_abs.sc:speed.sc 0.045 0.056
tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l. tmptrnd_bs.sc:c.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns. t-1:_. ts1:_.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
tmptrnd_bs.sc:np. tmptrnd_bs.sc:v. h_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-8.3520451 -0.3535263 0.0534660 0.5265006 6.6044778
Number of Observations: 43174
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
227 43174 Full model for Jaccard turnover, removing year 1
Try Bowler or Venter/Halpern human impact
summary(modTfullrem0)
Linear mixed-effects model fit by REML
Data: trends[i1, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.007404753 (Intr)
temptrend_abs.sc 0.012100255 0.189
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.003350251 1.507894
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.902506
Fixed effects: Jtutrendrem0 ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.613
REALMMarine -0.940 0.600
REALMTerrestrial -0.692 0.344 0.639
tsign1 -0.180 0.089 0.044 0.033
tempave.sc -0.027 0.010 0.037 0.010 -0.061
tempave_metab.sc 0.121 -0.064 -0.131 -0.086 -0.013 -0.573
seas.sc -0.130 0.091 0.183 -0.087 -0.105 0.243 -0.078
microclim.sc -0.098 0.105 0.112 -0.015 0.047 0.022 -0.072 0.146
mass.sc -0.015 0.056 0.026 0.072 0.021 0.001 -0.509 0.008
speed.sc 0.080 -0.035 -0.018 -0.071 -0.041 0.055 -0.224 0.052
lifespan.sc 0.079 -0.073 -0.090 -0.088 -0.041 -0.038 0.631 0.045
consumerfrac.sc -0.081 0.052 0.042 0.345 0.046 0.027 -0.042 -0.111
endothermfrac.sc -0.026 0.006 0.031 -0.191 0.020 0.506 -0.542 0.106
nspp.sc 0.033 -0.078 -0.101 -0.150 -0.043 0.017 0.010 -0.089
npp.sc 0.037 -0.012 -0.063 0.044 0.004 -0.244 0.243 -0.084
veg.sc -0.475 0.482 0.506 0.007 -0.002 0.129 -0.132 0.080
temptrend_abs.sc:REALMMarine 0.599 -0.964 -0.613 -0.321 -0.029 -0.031 0.069 -0.112
temptrend_abs.sc:REALMTerrestrial 0.353 -0.648 -0.329 -0.540 -0.017 0.007 0.044 0.134
temptrend_abs.sc:tsign1 0.107 -0.178 -0.050 -0.019 -0.519 0.079 -0.008 0.030
temptrend_abs.sc:tempave.sc 0.047 -0.021 -0.061 -0.005 0.062 -0.730 0.460 -0.202
temptrend_abs.sc:tempave_metab.sc -0.055 0.071 0.060 0.053 -0.018 0.446 -0.685 0.021
temptrend_abs.sc:seas.sc 0.077 -0.124 -0.105 0.124 0.067 -0.146 0.038 -0.732
temptrend_abs.sc:microclim.sc 0.088 -0.145 -0.094 0.017 -0.055 0.018 0.040 -0.081
temptrend_abs.sc:mass.sc 0.039 -0.051 -0.038 -0.062 -0.017 0.031 0.297 0.016
temptrend_abs.sc:speed.sc -0.058 0.057 0.034 0.080 0.034 -0.046 0.161 0.000
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.006
speed.sc 0.049 0.115
lifespan.sc 0.003 -0.848 -0.377
consumerfrac.sc -0.016 0.104 -0.336 -0.026
endothermfrac.sc 0.063 -0.096 -0.289 0.117 -0.078
nspp.sc -0.113 -0.164 0.018 0.181 -0.052 0.168
npp.sc -0.187 -0.036 0.043 0.033 -0.031 -0.217 -0.172
veg.sc 0.031 0.046 0.040 -0.055 -0.010 0.082 0.008 -0.273
temptrend_abs.sc:REALMMarine -0.107 -0.054 0.010 0.070 -0.034 -0.006 0.104 0.055
temptrend_abs.sc:REALMTerrestrial 0.021 -0.080 0.058 0.072 -0.187 0.091 0.148 -0.068
temptrend_abs.sc:tsign1 -0.046 -0.021 0.019 0.034 -0.027 0.013 0.031 -0.035
temptrend_abs.sc:tempave.sc 0.011 0.034 -0.020 -0.005 -0.010 -0.454 -0.029 0.176
temptrend_abs.sc:tempave_metab.sc -0.006 0.307 0.127 -0.382 0.066 0.406 -0.013 -0.177
temptrend_abs.sc:seas.sc -0.105 0.006 -0.017 -0.038 0.103 -0.073 0.069 0.065
temptrend_abs.sc:microclim.sc -0.812 0.006 -0.025 -0.005 0.000 -0.039 0.064 0.182
temptrend_abs.sc:mass.sc 0.024 -0.603 -0.058 0.524 -0.064 0.068 0.109 -0.012
temptrend_abs.sc:speed.sc -0.011 -0.067 -0.630 0.222 0.224 0.160 -0.065 -0.009
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.516
temptrend_abs.sc:REALMTerrestrial 0.021 0.600
temptrend_abs.sc:tsign1 -0.023 0.069 0.019
temptrend_abs.sc:tempave.sc -0.120 0.057 -0.035 -0.106
temptrend_abs.sc:tempave_metab.sc 0.102 -0.077 -0.080 0.021 -0.498
temptrend_abs.sc:seas.sc -0.074 0.164 -0.214 -0.051 0.285
temptrend_abs.sc:microclim.sc -0.051 0.152 -0.038 0.048 -0.028
temptrend_abs.sc:mass.sc -0.017 0.054 0.122 0.028 -0.051
temptrend_abs.sc:speed.sc -0.036 -0.017 -0.059 -0.017 0.064
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc -0.036
temptrend_abs.sc:microclim.sc 0.056 0.146
temptrend_abs.sc:mass.sc -0.539 -0.053 -0.049
temptrend_abs.sc:speed.sc -0.292 0.050 0.018
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.140
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-7.01782639 -0.26254472 -0.02396528 0.31237827 5.40866015
Number of Observations: 31072
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
218 31072 summary(modTfullfootprintrem0)Linear mixed-effects model fit by REML
Data: trends[i2, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.007176109 (Intr)
temptrend_abs.sc 0.015663961 0.105
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.003466799 1.508598
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.903419
Fixed effects: Jtutrendrem0 ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_footprint.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_.
temptrend_abs.sc -0.624
REALMMarine -0.940 0.606
REALMTerrestrial -0.673 0.357 0.615
tsign1 -0.179 0.082 0.041 0.041
tempave.sc -0.054 0.035 0.068 -0.001 -0.055
tempave_metab.sc 0.142 -0.078 -0.152 -0.076 -0.025 -0.572
seas.sc -0.130 0.085 0.182 -0.098 -0.107 0.247 -0.073
microclim.sc -0.091 0.083 0.105 -0.014 0.042 0.063 -0.091
mass.sc -0.015 0.053 0.023 0.082 0.022 0.014 -0.510
speed.sc 0.083 -0.041 -0.016 -0.091 -0.040 0.057 -0.216
lifespan.sc 0.083 -0.071 -0.096 -0.085 -0.043 -0.052 0.637
consumerfrac.sc -0.076 0.047 0.032 0.359 0.047 0.031 -0.037
endothermfrac.sc -0.046 0.018 0.050 -0.203 0.027 0.494 -0.546
nspp.sc 0.034 -0.071 -0.096 -0.150 -0.042 0.021 0.006
npp.sc 0.074 -0.038 -0.104 0.026 0.005 -0.280 0.264
veg.sc -0.505 0.473 0.536 0.013 -0.007 0.174 -0.164
temptrend_abs.sc:REALMMarine 0.607 -0.959 -0.620 -0.327 -0.023 -0.060 0.084
temptrend_abs.sc:REALMTerrestrial 0.350 -0.651 -0.319 -0.559 -0.023 0.018 0.032
temptrend_abs.sc:tsign1 0.101 -0.161 -0.041 -0.024 -0.518 0.078 -0.006
temptrend_abs.sc:tempave.sc 0.073 -0.058 -0.093 0.001 0.057 -0.736 0.448
temptrend_abs.sc:tempave_metab.sc -0.074 0.090 0.080 0.045 -0.009 0.440 -0.686
temptrend_abs.sc:seas.sc 0.080 -0.131 -0.112 0.127 0.070 -0.163 0.034
temptrend_abs.sc:microclim.sc 0.067 -0.107 -0.073 0.021 -0.051 -0.007 0.051
temptrend_abs.sc:mass.sc 0.036 -0.043 -0.036 -0.065 -0.018 0.028 0.299
temptrend_abs.sc:speed.sc -0.060 0.061 0.028 0.087 0.035 -0.053 0.150
ses.sc mcrcl. mss.sc spd.sc lfspn. cnsmr. endth.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc 0.156
mass.sc 0.010 -0.002
speed.sc 0.051 0.054 0.100
lifespan.sc 0.045 0.011 -0.837 -0.370
consumerfrac.sc -0.111 -0.011 0.121 -0.358 -0.010
endothermfrac.sc 0.104 0.088 -0.090 -0.281 0.103 -0.079
nspp.sc -0.100 -0.124 -0.168 0.024 0.179 -0.054 0.169
npp.sc -0.132 -0.206 -0.043 0.045 0.031 -0.058 -0.224
veg.sc 0.096 0.025 0.046 0.035 -0.053 -0.001 0.112
temptrend_abs.sc:REALMMarine -0.111 -0.088 -0.051 0.009 0.067 -0.025 -0.020
temptrend_abs.sc:REALMTerrestrial 0.138 0.024 -0.082 0.069 0.067 -0.207 0.112
temptrend_abs.sc:tsign1 0.040 -0.032 -0.019 0.017 0.033 -0.021 0.010
temptrend_abs.sc:tempave.sc -0.210 -0.016 0.025 -0.034 0.007 -0.008 -0.428
temptrend_abs.sc:tempave_metab.sc 0.015 0.006 0.308 0.120 -0.391 0.051 0.410
temptrend_abs.sc:seas.sc -0.736 -0.103 0.007 -0.031 -0.040 0.106 -0.068
temptrend_abs.sc:microclim.sc -0.081 -0.822 0.010 -0.034 -0.009 0.006 -0.051
temptrend_abs.sc:mass.sc 0.015 0.028 -0.604 -0.047 0.522 -0.068 0.064
temptrend_abs.sc:speed.sc -0.014 -0.017 -0.053 -0.636 0.211 0.243 0.162
nspp.s npp.sc veg.sc t_.:REALMM t_.:REALMT tm_.:1
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc -0.176
veg.sc 0.018 -0.310
temptrend_abs.sc:REALMMarine 0.098 0.081 -0.506
temptrend_abs.sc:REALMTerrestrial 0.138 -0.049 0.015 0.593
temptrend_abs.sc:tsign1 0.026 -0.036 -0.011 0.054 0.026
temptrend_abs.sc:tempave.sc -0.018 0.192 -0.170 0.097 -0.037 -0.102
temptrend_abs.sc:tempave_metab.sc -0.008 -0.190 0.129 -0.094 -0.067 0.018
temptrend_abs.sc:seas.sc 0.084 0.092 -0.092 0.177 -0.206 -0.064
temptrend_abs.sc:microclim.sc 0.072 0.194 -0.016 0.120 -0.046 0.024
temptrend_abs.sc:mass.sc 0.113 -0.009 -0.012 0.045 0.119 0.026
temptrend_abs.sc:speed.sc -0.062 -0.010 -0.033 -0.012 -0.065 -0.013
tmptrnd_bs.sc:t. tm_.:_. tmptrnd_bs.sc:ss.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc -0.473
temptrend_abs.sc:seas.sc 0.292 -0.025
temptrend_abs.sc:microclim.sc 0.014 0.048 0.161
temptrend_abs.sc:mass.sc -0.042 -0.542 -0.049
temptrend_abs.sc:speed.sc 0.075 -0.279 0.060
tmptrnd_bs.sc:mc. tmptrnd_bs.sc:ms.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc -0.057
temptrend_abs.sc:speed.sc 0.026 0.130
tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l. tmptrnd_bs.sc:c.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns. t-1:_. ts1:_.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
tmptrnd_bs.sc:np. tmptrnd_bs.sc:v. h_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-7.02311163 -0.26176466 -0.02363408 0.31203019 5.40892698
Number of Observations: 30801
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
210 30801 Full models for total and HM
i2 <- trends[, complete.cases(Jbetatrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
i3 <- trends[, complete.cases(Horntrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
randef <- list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
# full models
if(file.exists('temp/modTfullJbeta.rds')){
modTfullJbeta <- readRDS('temp/modTfullJbeta.rds')
} else {
modTfullJbeta <- lme(Jbetatrend ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_bowler.sc:REALM2,
random = randef, weights = varef, data = trends[i2,], method = 'REML')
saveRDS(modTfullJbeta, file = 'temp/modTfullJbeta.rds')
}
if(file.exists('temp/modTfullHorn.rds')){
modTfullHorn <- readRDS('temp/modTfullHorn.rds')
} else {
modTfullHorn <- lme(Horntrend ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_bowler.sc:REALM2,
random = randef, weights = varef, data = trends[i3,], method = 'REML')
saveRDS(modTfullHorn, file = 'temp/modTfullHorn.rds')
}
summary(modTfullJbeta)Linear mixed-effects model fit by REML
Data: trends[i2, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.06277468 (Intr)
temptrend_abs.sc 0.04847031 -0.175
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.0002039828 0.3158091
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.351927
Fixed effects: Jbetatrend ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.430
REALMMarine -0.929 0.402
REALMTerrestrial -0.876 0.377 0.815
tsign1 -0.026 0.017 0.006 0.009
tempave.sc -0.003 0.000 -0.004 -0.005 -0.070
tempave_metab.sc 0.038 -0.020 -0.030 -0.031 -0.022 -0.454
seas.sc -0.030 0.037 0.040 -0.012 -0.107 0.154 -0.084
microclim.sc -0.008 0.018 0.014 -0.006 0.020 -0.060 -0.022 0.127
mass.sc 0.012 0.006 -0.002 -0.002 0.019 0.021 -0.486 0.007
speed.sc 0.000 -0.005 0.009 0.017 -0.049 0.106 -0.277 0.102
lifespan.sc 0.024 -0.019 -0.019 -0.035 -0.015 -0.086 0.707 0.005
consumerfrac.sc -0.003 0.005 0.024 0.252 0.015 -0.044 -0.025 -0.051
endothermfrac.sc 0.133 -0.056 -0.068 -0.269 0.003 0.099 -0.111 0.012
nspp.sc -0.007 -0.011 -0.009 -0.019 -0.038 0.002 -0.023 -0.060
npp.sc 0.004 0.004 -0.005 0.006 0.013 -0.245 0.196 -0.088
veg.sc -0.075 0.119 0.073 -0.004 -0.016 0.141 -0.098 0.121
temptrend_abs.sc:REALMMarine 0.413 -0.960 -0.415 -0.359 -0.004 0.003 0.015 -0.046
temptrend_abs.sc:REALMTerrestrial 0.363 -0.858 -0.336 -0.433 -0.005 -0.011 0.020 0.026
temptrend_abs.sc:tsign1 0.013 -0.037 -0.005 -0.004 -0.484 0.050 -0.002 0.013
temptrend_abs.sc:tempave.sc 0.000 -0.005 -0.002 -0.002 0.052 -0.625 0.315 -0.052
temptrend_abs.sc:tempave_metab.sc -0.009 0.023 0.008 0.010 -0.012 0.335 -0.597 -0.010
temptrend_abs.sc:seas.sc 0.016 -0.050 -0.020 0.021 0.044 0.003 -0.008 -0.647
temptrend_abs.sc:microclim.sc 0.006 -0.026 -0.009 0.007 -0.044 0.089 -0.014 -0.053
temptrend_abs.sc:mass.sc 0.006 -0.010 -0.007 -0.006 -0.015 0.030 0.262 0.008
temptrend_abs.sc:speed.sc -0.004 0.018 -0.003 -0.001 0.026 -0.096 0.128 -0.035
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.056
speed.sc 0.072 0.129
lifespan.sc -0.027 -0.760 -0.455
consumerfrac.sc -0.001 0.014 -0.049 -0.047
endothermfrac.sc 0.009 0.001 -0.142 0.068 -0.328
nspp.sc -0.115 -0.208 0.014 0.134 -0.009 0.067
npp.sc -0.090 -0.044 0.048 0.035 -0.029 -0.049 -0.182
veg.sc -0.019 0.034 0.052 -0.029 -0.069 -0.012 0.002 -0.271
temptrend_abs.sc:REALMMarine -0.022 -0.010 -0.005 0.014 -0.010 0.036 0.020 0.008
temptrend_abs.sc:REALMTerrestrial 0.010 -0.008 -0.006 0.023 -0.100 0.135 0.031 -0.018
temptrend_abs.sc:tsign1 -0.041 -0.015 0.019 0.008 -0.004 0.006 0.025 -0.048
temptrend_abs.sc:tempave.sc 0.083 0.028 -0.064 0.031 -0.009 -0.087 -0.016 0.157
temptrend_abs.sc:tempave_metab.sc -0.034 0.244 0.069 -0.350 0.049 0.105 -0.006 -0.141
temptrend_abs.sc:seas.sc -0.088 0.007 -0.037 -0.005 0.041 0.019 0.040 0.061
temptrend_abs.sc:microclim.sc -0.693 -0.028 -0.029 0.018 0.002 0.009 0.045 0.103
temptrend_abs.sc:mass.sc -0.017 -0.555 -0.024 0.434 -0.005 0.014 0.131 0.005
temptrend_abs.sc:speed.sc -0.041 -0.044 -0.547 0.187 0.022 0.079 -0.024 0.006
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.129
temptrend_abs.sc:REALMTerrestrial 0.005 0.814
temptrend_abs.sc:tsign1 -0.004 0.012 0.007
temptrend_abs.sc:tempave.sc -0.140 -0.001 0.017 -0.076
temptrend_abs.sc:tempave_metab.sc 0.060 -0.019 -0.030 0.015 -0.420
temptrend_abs.sc:seas.sc -0.127 0.068 -0.052 -0.018 0.060
temptrend_abs.sc:microclim.sc 0.024 0.030 -0.023 0.047 -0.119
temptrend_abs.sc:mass.sc -0.005 0.016 0.018 0.026 -0.073
temptrend_abs.sc:speed.sc -0.057 0.001 0.000 -0.006 0.161
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc 0.031
temptrend_abs.sc:microclim.sc 0.104 0.139
temptrend_abs.sc:mass.sc -0.469 -0.045 -0.010
temptrend_abs.sc:speed.sc -0.223 0.074 0.046
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.063
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-7.3991424 -0.2444604 0.1291713 0.6077538 6.7430991
Number of Observations: 43493
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
235 43493 summary(modTfullHorn)Linear mixed-effects model fit by REML
Data: trends[i3, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.06157166 (Intr)
temptrend_abs.sc 0.04045651 -0.087
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.004171557 0.2960534
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.223338
Fixed effects: Horntrend ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.380
REALMMarine -0.917 0.352
REALMTerrestrial -0.858 0.326 0.781
tsign1 -0.032 0.020 0.007 0.013
tempave.sc -0.007 0.003 0.004 -0.005 -0.074
tempave_metab.sc 0.050 -0.023 -0.042 -0.039 -0.022 -0.506
seas.sc -0.043 0.037 0.059 -0.014 -0.114 0.205 -0.112
microclim.sc -0.017 0.022 0.025 -0.007 0.006 -0.034 -0.034 0.148
mass.sc 0.012 0.010 -0.001 0.003 0.021 0.023 -0.485 -0.004
speed.sc 0.005 -0.005 0.011 0.013 -0.049 0.105 -0.269 0.097
lifespan.sc 0.029 -0.020 -0.024 -0.040 -0.015 -0.087 0.691 0.018
consumerfrac.sc -0.004 0.007 0.002 0.268 0.019 -0.026 -0.021 -0.047
endothermfrac.sc 0.142 -0.048 -0.069 -0.299 0.004 0.153 -0.163 0.037
nspp.sc -0.010 -0.017 -0.009 -0.023 -0.031 0.016 -0.021 -0.019
npp.sc 0.015 0.004 -0.018 0.009 0.011 -0.292 0.228 -0.205
veg.sc -0.110 0.123 0.112 -0.003 -0.017 0.148 -0.106 0.135
temptrend_abs.sc:REALMMarine 0.364 -0.957 -0.362 -0.306 -0.006 -0.005 0.019 -0.048
temptrend_abs.sc:REALMTerrestrial 0.313 -0.840 -0.285 -0.387 -0.006 -0.008 0.025 0.041
temptrend_abs.sc:tsign1 0.016 -0.044 -0.006 -0.005 -0.489 0.042 0.003 0.015
temptrend_abs.sc:tempave.sc 0.004 -0.009 -0.008 -0.002 0.050 -0.571 0.306 -0.056
temptrend_abs.sc:tempave_metab.sc -0.014 0.028 0.013 0.016 -0.010 0.316 -0.574 -0.007
temptrend_abs.sc:seas.sc 0.018 -0.056 -0.025 0.029 0.045 -0.031 0.006 -0.606
temptrend_abs.sc:microclim.sc 0.009 -0.026 -0.012 0.007 -0.038 0.064 -0.005 -0.075
temptrend_abs.sc:mass.sc 0.007 -0.017 -0.008 -0.011 -0.013 0.021 0.260 0.005
temptrend_abs.sc:speed.sc -0.004 0.020 -0.005 -0.001 0.025 -0.081 0.123 -0.028
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.023
speed.sc 0.063 0.148
lifespan.sc 0.003 -0.778 -0.448
consumerfrac.sc -0.006 0.014 -0.062 -0.040
endothermfrac.sc 0.018 -0.011 -0.145 0.060 -0.317
nspp.sc -0.081 -0.222 0.000 0.149 -0.004 0.083
npp.sc -0.224 -0.034 0.046 0.028 -0.014 -0.083 -0.195
veg.sc 0.022 0.028 0.040 -0.022 -0.049 0.013 0.014 -0.276
temptrend_abs.sc:REALMMarine -0.025 -0.014 -0.007 0.015 -0.008 0.032 0.025 0.011
temptrend_abs.sc:REALMTerrestrial 0.012 -0.018 -0.004 0.027 -0.085 0.133 0.041 -0.024
temptrend_abs.sc:tsign1 -0.036 -0.013 0.018 0.007 -0.006 0.004 0.024 -0.046
temptrend_abs.sc:tempave.sc 0.076 0.025 -0.061 0.030 -0.011 -0.099 -0.017 0.144
temptrend_abs.sc:tempave_metab.sc -0.017 0.248 0.078 -0.347 0.050 0.120 -0.005 -0.142
temptrend_abs.sc:seas.sc -0.101 0.009 -0.035 -0.011 0.032 0.002 0.016 0.109
temptrend_abs.sc:microclim.sc -0.660 -0.008 -0.026 0.001 0.003 0.001 0.033 0.155
temptrend_abs.sc:mass.sc -0.008 -0.552 -0.034 0.435 -0.012 0.019 0.131 0.007
temptrend_abs.sc:speed.sc -0.034 -0.047 -0.533 0.186 0.020 0.084 -0.017 -0.001
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.136
temptrend_abs.sc:REALMTerrestrial 0.013 0.789
temptrend_abs.sc:tsign1 0.003 0.014 0.009
temptrend_abs.sc:tempave.sc -0.130 0.008 0.018 -0.068
temptrend_abs.sc:tempave_metab.sc 0.057 -0.021 -0.050 0.012 -0.434
temptrend_abs.sc:seas.sc -0.131 0.076 -0.080 -0.020 0.034
temptrend_abs.sc:microclim.sc 0.021 0.030 -0.029 0.050 -0.144
temptrend_abs.sc:mass.sc -0.001 0.021 0.033 0.025 -0.080
temptrend_abs.sc:speed.sc -0.047 0.004 -0.002 -0.004 0.162
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc 0.052
temptrend_abs.sc:microclim.sc 0.103 0.138
temptrend_abs.sc:mass.sc -0.465 -0.040 -0.008
temptrend_abs.sc:speed.sc -0.216 0.064 0.039
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.058
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-6.73851930 -0.38297739 0.03926292 0.54007605 6.19209050
Number of Observations: 42493
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
202 42493 Full models for total and HM (remove year 1)
i2 <- trends[, complete.cases(Jbetatrendrem0, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc,
veg.sc, human_bowler.sc)]
i3 <- trends[, complete.cases(Horntrendrem0, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend.sc, temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc,
veg.sc, human_bowler.sc)]
randef <- list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
# full models
if(file.exists('temp/modTfullJbetarem0.rds')){
modTfullJbetarem0 <- readRDS('temp/modTfullJbetarem0.rds')
} else {
modTfullJbetarem0 <- lme(Jbetatrendrem0 ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_bowler.sc:REALM2,
random = randef, weights = varef, data = trends[i2,], method = 'REML')
saveRDS(modTfullJbetarem0, file = 'temp/modTfullJbetarem0.rds')
}
if(file.exists('temp/modTfullHornrem0.rds')){
modTfullHornrem0 <- readRDS('temp/modTfullHornrem0.rds')
} else {
modTfullHornrem0 <- lme(Horntrendrem0 ~ temptrend_abs.sc*REALM +
temptrend_abs.sc*tsign +
temptrend_abs.sc*tempave.sc +
temptrend_abs.sc*tempave_metab.sc +
temptrend_abs.sc*seas.sc +
temptrend_abs.sc*microclim.sc +
temptrend_abs.sc*mass.sc +
temptrend_abs.sc*speed.sc +
temptrend_abs.sc*lifespan.sc +
temptrend_abs.sc*consumerfrac.sc +
temptrend_abs.sc*endothermfrac.sc +
temptrend_abs.sc*nspp.sc +
temptrend_abs.sc*thermal_bias.sc:tsign +
temptrend_abs.sc*npp.sc +
temptrend_abs.sc*veg.sc +
temptrend_abs.sc*human_bowler.sc:REALM2,
random = randef, weights = varef, data = trends[i3,], method = 'REML')
saveRDS(modTfullHornrem0, file = 'temp/modTfullHornrem0.rds')
}
summary(modTfullJbetarem0)Linear mixed-effects model fit by REML
Data: trends[i2, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.007723766 (Intr)
temptrend_abs.sc 0.017251054 0.308
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.002065633 0.8886333
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-1.821858
Fixed effects: Jbetatrendrem0 ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.478
REALMMarine -0.934 0.465
REALMTerrestrial -0.741 0.288 0.680
tsign1 -0.146 0.068 0.034 0.026
tempave.sc -0.017 0.007 0.026 0.004 -0.065
tempave_metab.sc 0.112 -0.049 -0.117 -0.071 -0.010 -0.547
seas.sc -0.110 0.070 0.160 -0.073 -0.106 0.226 -0.089
microclim.sc -0.080 0.087 0.098 -0.015 0.037 0.017 -0.069 0.169
mass.sc -0.007 0.052 0.022 0.050 0.018 0.006 -0.496 0.013
speed.sc 0.055 -0.033 0.000 -0.033 -0.049 0.067 -0.217 0.066
lifespan.sc 0.077 -0.060 -0.084 -0.079 -0.032 -0.048 0.639 0.030
consumerfrac.sc -0.071 0.042 0.040 0.330 0.042 0.014 -0.042 -0.100
endothermfrac.sc 0.029 -0.005 -0.009 -0.257 0.023 0.415 -0.481 0.093
nspp.sc 0.023 -0.070 -0.086 -0.114 -0.036 0.016 -0.007 -0.085
npp.sc 0.025 -0.006 -0.048 0.037 0.009 -0.246 0.231 -0.094
veg.sc -0.381 0.387 0.407 0.005 -0.005 0.139 -0.130 0.096
temptrend_abs.sc:REALMMarine 0.464 -0.954 -0.469 -0.267 -0.020 -0.026 0.050 -0.091
temptrend_abs.sc:REALMTerrestrial 0.283 -0.720 -0.262 -0.428 -0.011 0.000 0.026 0.119
temptrend_abs.sc:tsign1 0.082 -0.138 -0.035 -0.012 -0.510 0.078 -0.015 0.029
temptrend_abs.sc:tempave.sc 0.038 -0.019 -0.054 -0.007 0.062 -0.717 0.425 -0.187
temptrend_abs.sc:tempave_metab.sc -0.045 0.059 0.048 0.037 -0.019 0.425 -0.673 0.028
temptrend_abs.sc:seas.sc 0.068 -0.098 -0.097 0.099 0.064 -0.129 0.041 -0.738
temptrend_abs.sc:microclim.sc 0.072 -0.116 -0.083 0.019 -0.046 0.023 0.035 -0.104
temptrend_abs.sc:mass.sc 0.033 -0.041 -0.035 -0.047 -0.016 0.030 0.289 0.010
temptrend_abs.sc:speed.sc -0.051 0.053 0.023 0.056 0.037 -0.060 0.131 -0.016
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.015
speed.sc 0.062 0.112
lifespan.sc -0.003 -0.825 -0.385
consumerfrac.sc -0.023 0.069 -0.267 -0.032
endothermfrac.sc 0.049 -0.088 -0.300 0.111 -0.140
nspp.sc -0.122 -0.171 0.015 0.167 -0.033 0.165
npp.sc -0.189 -0.032 0.046 0.029 -0.034 -0.184 -0.168
veg.sc 0.032 0.046 0.048 -0.054 -0.017 0.065 0.008 -0.286
temptrend_abs.sc:REALMMarine -0.092 -0.053 0.004 0.058 -0.025 0.011 0.096 0.047
temptrend_abs.sc:REALMTerrestrial 0.026 -0.060 0.040 0.051 -0.152 0.098 0.114 -0.055
temptrend_abs.sc:tsign1 -0.033 -0.017 0.018 0.024 -0.021 0.011 0.022 -0.043
temptrend_abs.sc:tempave.sc 0.013 0.028 -0.033 0.006 -0.010 -0.354 -0.023 0.183
temptrend_abs.sc:tempave_metab.sc -0.006 0.293 0.089 -0.371 0.066 0.370 -0.006 -0.172
temptrend_abs.sc:seas.sc -0.125 0.002 -0.026 -0.027 0.089 -0.059 0.069 0.083
temptrend_abs.sc:microclim.sc -0.813 -0.004 -0.036 0.002 0.009 -0.025 0.074 0.192
temptrend_abs.sc:mass.sc 0.017 -0.596 -0.044 0.507 -0.051 0.058 0.114 -0.015
temptrend_abs.sc:speed.sc -0.030 -0.061 -0.608 0.192 0.182 0.164 -0.045 0.001
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.417
temptrend_abs.sc:REALMTerrestrial 0.019 0.665
temptrend_abs.sc:tsign1 -0.021 0.049 0.014
temptrend_abs.sc:tempave.sc -0.133 0.045 -0.012 -0.101
temptrend_abs.sc:tempave_metab.sc 0.095 -0.060 -0.056 0.029 -0.454
temptrend_abs.sc:seas.sc -0.095 0.138 -0.175 -0.045 0.263
temptrend_abs.sc:microclim.sc -0.052 0.126 -0.043 0.031 -0.023
temptrend_abs.sc:mass.sc -0.014 0.049 0.086 0.021 -0.045
temptrend_abs.sc:speed.sc -0.042 -0.007 -0.034 -0.014 0.090
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc -0.042
temptrend_abs.sc:microclim.sc 0.058 0.170
temptrend_abs.sc:mass.sc -0.521 -0.043 -0.039
temptrend_abs.sc:speed.sc -0.261 0.070 0.040
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.125
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-8.38808782 -0.33122885 -0.03014344 0.33937884 8.28289240
Number of Observations: 31072
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
218 31072 summary(modTfullHornrem0)Linear mixed-effects model fit by REML
Data: trends[i3, ]
Random effects:
Formula: ~temptrend_abs.sc | STUDY_ID
Structure: General positive-definite, Log-Cholesky parametrization
StdDev Corr
(Intercept) 0.01569909 (Intr)
temptrend_abs.sc 0.02631579 0.237
Formula: ~1 | rarefyID %in% STUDY_ID
(Intercept) Residual
StdDev: 0.01343706 2.039905
Variance function:
Structure: Power of variance covariate
Formula: ~nyrBT
Parameter estimates:
power
-2.179888
Fixed effects: Horntrendrem0 ~ temptrend_abs.sc * REALM + temptrend_abs.sc * tsign + temptrend_abs.sc * tempave.sc + temptrend_abs.sc * tempave_metab.sc + temptrend_abs.sc * seas.sc + temptrend_abs.sc * microclim.sc + temptrend_abs.sc * mass.sc + temptrend_abs.sc * speed.sc + temptrend_abs.sc * lifespan.sc + temptrend_abs.sc * consumerfrac.sc + temptrend_abs.sc * endothermfrac.sc + temptrend_abs.sc * nspp.sc + temptrend_abs.sc * thermal_bias.sc:tsign + temptrend_abs.sc * npp.sc + temptrend_abs.sc * veg.sc + temptrend_abs.sc * human_bowler.sc:REALM2 Correlation:
(Intr) tmpt_. REALMM REALMT tsign1 tmpv.s tmpv_. ses.sc
temptrend_abs.sc -0.438
REALMMarine -0.942 0.425
REALMTerrestrial -0.717 0.272 0.656
tsign1 -0.095 0.060 0.014 0.023
tempave.sc -0.036 0.014 0.042 0.012 -0.061
tempave_metab.sc 0.103 -0.038 -0.100 -0.082 -0.013 -0.662
seas.sc -0.121 0.053 0.165 -0.099 -0.110 0.213 -0.074
microclim.sc -0.059 0.048 0.071 -0.017 0.031 -0.034 -0.041 0.086
mass.sc -0.011 0.037 0.026 0.054 0.018 0.012 -0.444 0.018
speed.sc 0.043 -0.028 0.007 -0.004 -0.052 0.097 -0.214 0.063
lifespan.sc 0.066 -0.045 -0.071 -0.091 -0.028 -0.068 0.589 0.034
consumerfrac.sc -0.039 0.034 0.014 0.296 0.046 0.028 -0.063 -0.117
endothermfrac.sc 0.023 -0.004 -0.009 -0.236 0.021 0.518 -0.551 0.084
nspp.sc 0.011 -0.066 -0.061 -0.107 -0.007 0.042 -0.014 -0.020
npp.sc 0.032 0.013 -0.047 0.061 -0.005 -0.240 0.209 -0.205
veg.sc -0.477 0.315 0.495 0.019 -0.012 0.128 -0.102 0.111
temptrend_abs.sc:REALMMarine 0.427 -0.953 -0.430 -0.254 -0.020 -0.026 0.035 -0.066
temptrend_abs.sc:REALMTerrestrial 0.248 -0.695 -0.230 -0.416 -0.007 -0.007 0.038 0.147
temptrend_abs.sc:tsign1 0.055 -0.120 -0.018 -0.002 -0.517 0.056 -0.013 0.012
temptrend_abs.sc:tempave.sc 0.041 -0.032 -0.053 -0.020 0.055 -0.610 0.405 -0.103
temptrend_abs.sc:tempave_metab.sc -0.040 0.042 0.037 0.044 -0.016 0.396 -0.626 -0.013
temptrend_abs.sc:seas.sc 0.060 -0.096 -0.082 0.125 0.057 -0.096 0.017 -0.661
temptrend_abs.sc:microclim.sc 0.026 -0.062 -0.033 0.015 -0.052 0.052 0.003 -0.032
temptrend_abs.sc:mass.sc 0.026 -0.038 -0.028 -0.039 -0.011 0.028 0.250 -0.003
temptrend_abs.sc:speed.sc -0.026 0.046 0.004 0.032 0.034 -0.054 0.118 -0.004
mcrcl. mss.sc spd.sc lfspn. cnsmr. endth. nspp.s npp.sc
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc 0.013
speed.sc 0.058 0.153
lifespan.sc 0.007 -0.826 -0.394
consumerfrac.sc -0.021 0.040 -0.192 -0.052
endothermfrac.sc 0.030 -0.080 -0.254 0.082 -0.140
nspp.sc -0.069 -0.191 0.010 0.181 -0.017 0.162
npp.sc -0.244 -0.032 0.036 0.026 -0.001 -0.170 -0.200
veg.sc 0.008 0.020 0.024 -0.022 -0.008 0.077 0.019 -0.205
temptrend_abs.sc:REALMMarine -0.054 -0.038 -0.001 0.042 -0.024 0.014 0.084 0.016
temptrend_abs.sc:REALMTerrestrial 0.027 -0.055 0.024 0.053 -0.133 0.087 0.111 -0.076
temptrend_abs.sc:tsign1 -0.029 -0.014 0.021 0.018 -0.014 0.004 0.011 -0.034
temptrend_abs.sc:tempave.sc 0.058 0.035 -0.050 0.014 -0.031 -0.323 -0.030 0.139
temptrend_abs.sc:tempave_metab.sc -0.014 0.272 0.106 -0.359 0.069 0.354 0.007 -0.142
temptrend_abs.sc:seas.sc -0.074 -0.002 -0.026 -0.028 0.096 -0.041 0.035 0.126
temptrend_abs.sc:microclim.sc -0.723 0.006 -0.030 -0.008 0.005 -0.001 0.044 0.181
temptrend_abs.sc:mass.sc 0.006 -0.592 -0.064 0.495 -0.032 0.051 0.108 -0.011
temptrend_abs.sc:speed.sc -0.029 -0.071 -0.584 0.197 0.124 0.134 -0.048 -0.011
veg.sc t_.:REALMM t_.:REALMT tm_.:1 tmptrnd_bs.sc:t.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine -0.335
temptrend_abs.sc:REALMTerrestrial 0.027 0.639
temptrend_abs.sc:tsign1 -0.007 0.041 0.005
temptrend_abs.sc:tempave.sc -0.111 0.050 0.024 -0.107
temptrend_abs.sc:tempave_metab.sc 0.066 -0.035 -0.081 0.036 -0.503
temptrend_abs.sc:seas.sc -0.102 0.130 -0.222 -0.034 0.156
temptrend_abs.sc:microclim.sc 0.020 0.073 -0.048 0.027 -0.108
temptrend_abs.sc:mass.sc -0.003 0.043 0.083 0.023 -0.073
temptrend_abs.sc:speed.sc -0.032 0.003 -0.024 -0.010 0.106
tm_.:_. tmptrnd_bs.sc:ss. tmptrnd_bs.sc:mc.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc 0.012
temptrend_abs.sc:microclim.sc 0.092 0.098
temptrend_abs.sc:mass.sc -0.471 -0.025 -0.024
temptrend_abs.sc:speed.sc -0.253 0.062 0.033
tmptrnd_bs.sc:ms. tmptrnd_bs.sc:sp. tmptrnd_bs.sc:l.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc 0.125
tmptrnd_bs.sc:c. tmptrnd_bs.sc:nd. tmptrnd_bs.sc:ns.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
t-1:_. ts1:_. tmptrnd_bs.sc:np. tmptrnd_bs.sc:v.
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
h_.:REALM2T h_.:REALM2M t_.:-1 t_.:1: t_.:_.:REALM2T
temptrend_abs.sc
REALMMarine
REALMTerrestrial
tsign1
tempave.sc
tempave_metab.sc
seas.sc
microclim.sc
mass.sc
speed.sc
lifespan.sc
consumerfrac.sc
endothermfrac.sc
nspp.sc
npp.sc
veg.sc
temptrend_abs.sc:REALMMarine
temptrend_abs.sc:REALMTerrestrial
temptrend_abs.sc:tsign1
temptrend_abs.sc:tempave.sc
temptrend_abs.sc:tempave_metab.sc
temptrend_abs.sc:seas.sc
temptrend_abs.sc:microclim.sc
temptrend_abs.sc:mass.sc
temptrend_abs.sc:speed.sc
[ reached getOption("max.print") -- omitted 14 rows ]
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-5.79037325 -0.27332859 -0.02733247 0.28919907 5.91687491
Number of Observations: 30392
Number of Groups:
STUDY_ID rarefyID %in% STUDY_ID
190 30392 Plots from the full models
Plot the coefficients
Plot interactions (Jaccard turnover)
# set up the interactions to plot
ints <- data.frame(vars = c('tsign', 'tempave', 'tempave_metab', 'seas', 'microclim', 'mass', 'speed', 'lifespan', 'consumerfrac', 'endothermfrac', 'nspp', 'thermal_bias', 'npp', 'veg', 'human_bowler', 'human_bowler'),
min = c(1, -10, 10, 0.1, 0, 0, 0, 0.3, 0, 0, 0.3, -10, 1.9, 0, 0, 0),
max = c(2, 30, 40, 16, 6, 8, 2, 2, 1, 1, 2.6, 10, 3.7, 1, 9, 9),
log = c(F, F, F, F, F, T, T, T, F, F, T, F, T, F, F, F),
len = c(2, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100),
discrete = c(T, F, F, F, F, F, F, F, F, F, F, F, F, F, F, F),
REALM = c(rep('Freshwater', 15), 'Marine'),
REALM2 = c(rep('TerrFresh', 15), 'Marine'),
stringsAsFactors = FALSE)
basetab <- data.frame(tempave.sc = 0, tempave_metab.sc = 0,
seas.sc = 0, microclim.sc = 0, mass.sc = 0,
speed.sc = 0, lifespan.sc = 0, endothermfrac.sc = 0,
nspp.sc = 0, thermal_bias.sc = 0, npp.sc = 0, human_bowler.sc = 0, veg.sc = 0,
consumerfrac.sc = 0,
nyrBT = 20, STUDY_ID = 127L, rarefyID = '127_514668')
# make the data frames for each interaction to plot
for(j in 1:nrow(ints)){
# set up a grid of temperature trends and the interacting variable
if(ints$log[j]) intvars <- list(temptrend = seq(-1.5, 1.5, length.out = 100),
new = 10^seq(ints$min[j], ints$max[j], length.out = ints$len[j]),
var = ints$vars[j])
if(!ints$log[j]) intvars <- list(temptrend = seq(-1.5, 1.5, length.out = 100),
new = seq(ints$min[j], ints$max[j], length.out = ints$len[j]),
var = ints$vars[j])
names(intvars) <- c('temptrend', ints$vars[j], 'var')
thisdat <- expand.grid(intvars)
# scale the interacting variable
cent <- attr(trends[[paste0(ints$var[j], '.sc')]], 'scaled:center')
scl <- attr(trends[[paste0(ints$var[j], '.sc')]], 'scaled:scale')
if(!is.null(cent) & !is.null(scl)){
if(ints$log[j]) thisdat[[paste0(ints$var[j], '.sc')]] <- (log(thisdat[[ints$var[j]]]) - cent)/scl
if(!ints$log[j]) thisdat[[paste0(ints$var[j], '.sc')]] <- (thisdat[[ints$var[j]]] - cent)/scl
}
# merge with the rest of the columns
if(ints$var[j] != 'tsign') colnamestouse <- setdiff(colnames(basetab), paste0(ints$var[j], '.sc'))
if(ints$var[j] == 'tsign') colnamestouse <- setdiff(colnames(basetab), ints$var[j])
thisdat <- cbind(thisdat, basetab[, colnamestouse])
# add realm
thisdat$REALM <- ints$REALM[j]
thisdat$REALM2 <- ints$REALM2[j]
# merge with the previous iterations
if(j == 1) newdat <- thisdat
if(j > 1){
colstoadd <- setdiff(colnames(thisdat), colnames(newdat))
for(toadd in colstoadd){
newdat[[toadd]] <- NA
}
colstoadd2 <- setdiff(colnames(newdat), colnames(thisdat))
for(toadd in colstoadd2){
thisdat[[toadd]] <- NA
}
newdat <- rbind(newdat, thisdat)
}
}
# character so that new levels can be added
newdat$REALM <- as.character(newdat$REALM)
newdat$REALM2 <- as.character(newdat$REALM2)
# add extra rows so that all factor levels are represented (for predict.lme to work)
newdat <- rbind(newdat[1:4, ], newdat)
newdat$REALM[1:4] <- c('Marine', 'Marine', 'Terrestrial', 'Terrestrial')
newdat$REALM2[1:4] <- c('Marine', 'Marine', 'TerrFresh', 'TerrFresh')
newdat$temptrend[1:4] <- c(-1, 1, -1, 1)
# trim to at least some temperature change (so that tsign is -1 or 1)
newdat <- newdat[newdat$temptrend != 0,]
# scale the temperature vars
newdat$temptrend.sc <- newdat$temptrend/attr(trends$temptrend.sc, 'scaled:scale')
newdat$temptrend_abs <- abs(newdat$temptrend)
newdat$temptrend_abs.sc <- (newdat$temptrend_abs)/attr(trends$temptrend_abs.sc, 'scaled:scale')
newdat$tsign <- factor(sign(newdat$temptrend))
# make predictions
newdat$preds <- predict(object = modTfull1, newdata = newdat, level = 0)
#remove the extra rows
newdat <- newdat[5:nrow(newdat), ]
# prep the plots
intplots <- vector('list', nrow(ints))
for(j in 1:length(intplots)){
subs <- newdat$var == ints$vars[j] & newdat$temptrend > 0 # select warming side
xvar <- 'temptrend_abs'
title <- ints$vars[j]
if(ints$vars[j] %in% c('tsign')){
subs <- newdat$var == ints$vars[j]
}
if(ints$vars[j] %in% c('thermal_bias')){
subs <- newdat$var == ints$vars[j]
xvar <- 'temptrend'
}
if(ints$vars[j] %in% c('human_bowler')){
subs <- newdat$var == ints$vars[j] & newdat$temptrend > 0 & newdat$REALM2 == ints$REALM2[j]
title <- paste0('human:', ints$REALM2[j])
}
thisplot <- ggplot(newdat[subs, ],
aes_string(x = xvar, y = 'preds',
group = ints$vars[j],
color = ints$vars[j])) +
geom_line() +
coord_cartesian(ylim = c(0, 1)) +
theme(plot.margin = unit(c(0.5,0,0.5,0), 'cm')) +
labs(title = title)
if(ints$log[j] & !ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_distiller(palette = "YlGnBu", trans = 'log')
}
if(!ints$log[j] & !ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_distiller(palette = "YlGnBu", trans = 'identity')
}
if(ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_brewer(palette = "Dark2")
}
}
#grid.arrange(grobs = intplots, '+', theme(plot.margin = unit(c(0,0,0,0), 'cm'))), ncol=2)
#do.call('grid.arrange', c(intplots, ncol = 2))
grid.arrange(grobs = intplots, ncol = 3)
Positive thermal bias means that the species in the assemblage have warmer thermal niches than the environment (and vice versa).
Plot interactions (Jaccard turnover) without year 1
# set up the interactions to plot
ints <- data.frame(vars = c('tsign', 'tempave', 'tempave_metab', 'seas', 'microclim', 'mass', 'speed', 'lifespan', 'consumerfrac', 'endothermfrac', 'nspp', 'thermal_bias', 'npp', 'veg', 'human_bowler', 'human_bowler'),
min = c(1, -10, 10, 0.1, 0, 0, 0, 0.3, 0, 0, 0.3, -10, 1.9, 0, 0, 0),
max = c(2, 30, 40, 16, 6, 8, 2, 2, 1, 1, 2.6, 10, 3.7, 1, 9, 9),
log = c(F, F, F, F, F, T, T, T, F, F, T, F, T, F, F, F),
len = c(2, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100),
discrete = c(T, F, F, F, F, F, F, F, F, F, F, F, F, F, F, F),
REALM = c(rep('Freshwater', 15), 'Marine'),
REALM2 = c(rep('TerrFresh', 15), 'Marine'),
stringsAsFactors = FALSE)
basetab <- data.frame(tempave.sc = 0, tempave_metab.sc = 0,
seas.sc = 0, microclim.sc = 0, mass.sc = 0,
speed.sc = 0, lifespan.sc = 0, endothermfrac.sc = 0,
nspp.sc = 0, thermal_bias.sc = 0, npp.sc = 0, human_bowler.sc = 0, veg.sc = 0,
consumerfrac.sc = 0,
nyrBT = 20, STUDY_ID = 127L, rarefyID = '127_514668')
# make the data frames for each interaction to plot
for(j in 1:nrow(ints)){
# set up a grid of temperature trends and the interacting variable
if(ints$log[j]) intvars <- list(temptrend = seq(-1.5, 1.5, length.out = 100),
new = 10^seq(ints$min[j], ints$max[j], length.out = ints$len[j]),
var = ints$vars[j])
if(!ints$log[j]) intvars <- list(temptrend = seq(-1.5, 1.5, length.out = 100),
new = seq(ints$min[j], ints$max[j], length.out = ints$len[j]),
var = ints$vars[j])
names(intvars) <- c('temptrend', ints$vars[j], 'var')
thisdat <- expand.grid(intvars)
# scale the interacting variable
cent <- attr(trends[[paste0(ints$var[j], '.sc')]], 'scaled:center')
scl <- attr(trends[[paste0(ints$var[j], '.sc')]], 'scaled:scale')
if(!is.null(cent) & !is.null(scl)){
if(ints$log[j]) thisdat[[paste0(ints$var[j], '.sc')]] <- (log(thisdat[[ints$var[j]]]) - cent)/scl
if(!ints$log[j]) thisdat[[paste0(ints$var[j], '.sc')]] <- (thisdat[[ints$var[j]]] - cent)/scl
}
# merge with the rest of the columns
if(ints$var[j] != 'tsign') colnamestouse <- setdiff(colnames(basetab), paste0(ints$var[j], '.sc'))
if(ints$var[j] == 'tsign') colnamestouse <- setdiff(colnames(basetab), ints$var[j])
thisdat <- cbind(thisdat, basetab[, colnamestouse])
# add realm
thisdat$REALM <- ints$REALM[j]
thisdat$REALM2 <- ints$REALM2[j]
# merge with the previous iterations
if(j == 1) newdat <- thisdat
if(j > 1){
colstoadd <- setdiff(colnames(thisdat), colnames(newdat))
for(toadd in colstoadd){
newdat[[toadd]] <- NA
}
colstoadd2 <- setdiff(colnames(newdat), colnames(thisdat))
for(toadd in colstoadd2){
thisdat[[toadd]] <- NA
}
newdat <- rbind(newdat, thisdat)
}
}
# character so that new levels can be added
newdat$REALM <- as.character(newdat$REALM)
newdat$REALM2 <- as.character(newdat$REALM2)
# add extra rows so that all factor levels are represented (for predict.lme to work)
newdat <- rbind(newdat[1:4, ], newdat)
newdat$REALM[1:4] <- c('Marine', 'Marine', 'Terrestrial', 'Terrestrial')
newdat$REALM2[1:4] <- c('Marine', 'Marine', 'TerrFresh', 'TerrFresh')
newdat$temptrend[1:4] <- c(-1, 1, -1, 1)
# trim to at least some temperature change (so that tsign is -1 or 1)
newdat <- newdat[newdat$temptrend != 0,]
# scale the temperature vars
newdat$temptrend.sc <- newdat$temptrend/attr(trends$temptrend.sc, 'scaled:scale')
newdat$temptrend_abs <- abs(newdat$temptrend)
newdat$temptrend_abs.sc <- (newdat$temptrend_abs)/attr(trends$temptrend_abs.sc, 'scaled:scale')
newdat$tsign <- factor(sign(newdat$temptrend))
# make predictions
newdat$preds <- predict(object = modTfullrem0, newdata = newdat, level = 0)
#remove the extra rows
newdat <- newdat[5:nrow(newdat), ]
# prep the plots
intplots <- vector('list', nrow(ints))
for(j in 1:length(intplots)){
subs <- newdat$var == ints$vars[j] & newdat$temptrend > 0 # select warming side
xvar <- 'temptrend_abs'
title <- ints$vars[j]
if(ints$vars[j] %in% c('tsign')){
subs <- newdat$var == ints$vars[j]
}
if(ints$vars[j] %in% c('thermal_bias')){
subs <- newdat$var == ints$vars[j]
xvar <- 'temptrend'
}
if(ints$vars[j] %in% c('human_bowler')){
subs <- newdat$var == ints$vars[j] & newdat$temptrend > 0 & newdat$REALM2 == ints$REALM2[j]
title <- paste0('human:', ints$REALM2[j])
}
thisplot <- ggplot(newdat[subs, ],
aes_string(x = xvar, y = 'preds',
group = ints$vars[j],
color = ints$vars[j])) +
geom_line() +
coord_cartesian(ylim = c(-0.6, 0.6)) +
theme(plot.margin = unit(c(0.5,0,0.5,0), 'cm')) +
labs(title = title)
if(ints$log[j] & !ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_distiller(palette = "YlGnBu", trans = 'log')
}
if(!ints$log[j] & !ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_distiller(palette = "YlGnBu", trans = 'identity')
}
if(ints$discrete[j]){
intplots[[j]] <- thisplot + scale_color_brewer(palette = "Dark2")
}
}
#grid.arrange(grobs = intplots, '+', theme(plot.margin = unit(c(0,0,0,0), 'cm'))), ncol=2)
#do.call('grid.arrange', c(intplots, ncol = 2))
grid.arrange(grobs = intplots, ncol = 3)
Plot residuals against each predictor (Jaccard turnover)
resids <- resid(modTfull1)
preds <- getData(modTfull1)
col = '#00000033'
cex = 0.5
par(mfrow = c(5,4))
boxplot(resids ~ preds$REALM, cex = cex, col = col)
plot(preds$temptrend_abs.sc, resids, cex = cex, col = col)plot(preds$tsign, resids, cex = cex, col = col)
plot(preds$tempave.sc, resids, cex = cex, col = col)plot(preds$tempave_metab.sc, resids, cex = cex, col = col)
plot(preds$seas.sc, resids, cex = cex, col = col)plot(preds$microclim.sc, resids, cex = cex, col = col)
plot(preds$mass.sc, resids, cex = cex, col = col)plot(preds$speed.sc, resids, cex = cex, col = col)
plot(preds$lifespan.sc, resids, cex = cex, col = col)plot(preds$consumerfrac.sc, resids, cex = cex, col = col)
plot(preds$endothermfrac.sc, resids, cex = cex, col = col)plot(preds$nspp.sc, resids, cex = cex, col = col)
plot(preds$thermal_bias.sc, resids, cex = cex, col = col)plot(preds$npp.sc, resids, cex = cex, col = col)
plot(preds$veg.sc, resids, cex = cex, col = col)plot(preds$human_bowler.sc, resids, cex = cex, col = col)
Remove each term from the full model
Jaccard turnover
Jbeta and Horn models
if(file.exists('output/aics_from_full.csv')){
aicsfromfull <- read.csv('output/aics_from_full.csv')
if(all(c('dAIC_Jbeta', 'dAIC_Horn') %in% colnames(aicsfromfull))){
runJbetaHorn <- FALSE
} else {
runJbetaHorn <- TRUE
}
} else {
runJbetaHorn <- TRUE
}
if(runJbetaHorn){
i <- trends[, complete.cases(Jbetatrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
i2 <- trends[, complete.cases(Horntrend, REALM, tempave.sc, tempave_metab.sc, seas.sc, microclim.sc,
temptrend_abs.sc, mass.sc, speed.sc, lifespan.sc,
consumerfrac.sc, endothermfrac.sc, nspp.sc, thermal_bias.sc, npp.sc, veg.sc, human_bowler.sc)]
randef <- list(STUDY_ID = ~ temptrend_abs.sc, rarefyID = ~1)
varef <- varPower(-0.5, ~nyrBT)
terms <- c('temptrend_abs.sc*REALM',
'temptrend_abs.sc*tsign',
'temptrend_abs.sc*tempave.sc',
'temptrend_abs.sc*tempave_metab.sc',
'temptrend_abs.sc*seas.sc',
'temptrend_abs.sc*microclim.sc',
'temptrend_abs.sc*mass.sc',
'temptrend_abs.sc*speed.sc',
'temptrend_abs.sc*lifespan.sc',
'temptrend_abs.sc*consumerfrac.sc',
'temptrend_abs.sc*endothermfrac.sc',
'temptrend_abs.sc*nspp.sc',
'temptrend_abs.sc*thermal_bias.sc:tsign',
'temptrend_abs.sc*npp.sc',
'temptrend_abs.sc*veg.sc',
'temptrend_abs.sc*human_bowler.sc:REALM2')
modTJbetadrops <- vector('list', length(terms)+2)
modTHorndrops <- vector('list', length(terms)+2)
names(modTJbetadrops) <- c('full', '-temptrend_abs.sc', paste0('-', terms))
names(modTHorndrops) <- c('full', '-temptrend_abs.sc', paste0('-', terms))
# fit full model with ML for model comparison
modTJbetadrops[[1]] <- lme(formula(paste0('Jbetatrend ~ ', paste(terms, collapse = ' + '))),
random = randef, weights = varef, data = trends[i,], method = 'ML')
modTHorndrops[[1]] <- lme(formula(paste0('Horntrend ~ ', paste(terms, collapse = ' + '))),
random = randef, weights = varef, data = trends[i2,], method = 'ML')
# w/out temptrend
modTJbetadrops[[2]] <- lme(formula(paste0('Jbetatrend ~ ', paste(gsub('temptrend_abs.sc\\*', '', terms), collapse = ' + '))),
random = list(STUDY_ID = ~ 1, rarefyID = ~1), weights = varef, data = trends[i,], method = 'ML')
modTHorndrops[[2]] <- lme(formula(paste0('Horntrend ~ ', paste(gsub('temptrend_abs.sc\\*', '', terms), collapse = ' + '))),
random = list(STUDY_ID = ~ 1, rarefyID = ~1), weights = varef, data = trends[i2,], method = 'ML')
for(j in 1:length(terms)){
print(j)
modTJbetadrops[[j+2]] <- lme(formula(paste0('Jbetatrend ~ ', paste(terms[-j], collapse = ' + '))),
random = randef, weights = varef, data = trends[i,], method = 'ML',
control = lmeContrl(returnObject = TRUE)) # return fitted object even if a convergence error
modTHorndrops[[j+2]] <- lme(formula(paste0('Horntrend ~ ', paste(terms[-j], collapse = ' + '))),
random = randef, weights = varef, data = trends[i2,], method = 'ML')
}
aicsJbeta <- sapply(modTJbetadrops, AIC)
aicsHorn <- sapply(modTHorndrops, AIC)
if(exists('aicsfromfull')){
aicsfromfull$dAIC_Jbeta <- aicsJbeta - aicsJbeta[1]
aicsfromfull$dAIC_Horn <- aicsHorn - aicsHorn[1]
} else {
aicsfromfull <- data.frame(mod = names(aics),
dAIC_Jbeta = aicsJbeta - aicsJbeta[1],
dAIC_Horn <- aicsHorn - aicsHorn[1])
}
write.csv(aicsfromfull, file = 'output/aics_from_full.csv', row.names = FALSE)
}
aicsfromfullPlot deltaAICs for all 3 models
# transform for a plot
aicsfromfulllong <- reshape(aicsfromfull, direction = 'long',
varying = c('dAIC_Jtu', 'dAIC_Jbeta', 'dAIC_Horn'),
v.names = 'dAIC',
idvar = 'mod',
timevar = 'type',
times = c('Jtu', 'Jbeta', 'Horn'))
trans = function(x) sign(x)*sqrt(abs(x))
aicsfromfulllong$dAIC_tr <- trans(aicsfromfulllong$dAIC)
# plot
xlims <- range(aicsfromfulllong$dAIC_tr)
xticks <- c(-10, 0, 10, 100, 1000, 10000)
par(mai = c(0.5, 3, 0.1, 0.1))
with(aicsfromfulllong[aicsfromfulllong$type == 'Jtu',], plot(dAIC_tr, nrow(aicsfromfull):1,
col = 'light grey', xlim = xlims, yaxt = 'n', ylab = '', xaxt = 'n'))
with(aicsfromfulllong[aicsfromfulllong$type == 'Jbeta',], points(dAIC_tr, nrow(aicsfromfull):1 - 0.1, col = 'dark grey'))
with(aicsfromfulllong[aicsfromfulllong$type == 'Horn',], points(dAIC_tr, nrow(aicsfromfull):1 - 0.2, col = 'black'))
axis(2, at = nrow(aicsfromfull):1, labels = aicsfromfull$mod, las = 1, cex.axis = 0.7)
axis(1, at = trans(xticks), labels = xticks, cex.axis = 0.5)
abline(v = 0, lty =2, col = 'grey')Light grey is for Jaccard turnover, dark grey is for Jaccard total, black is for Morisita-Horn
Plot interaction coefficients from all full models
# fig.width = 3, fig.height = 5, out.height=2.5, out.width=3, fig.retina =3 for macbook screen
# double that for external monitor
coefs <- as.data.table(summary(modTfull1)$tTable)
coefs2 <- as.data.table(summary(modTfullJbeta)$tTable)
coefs3 <- as.data.table(summary(modTfullHorn)$tTable)
coefs4 <- as.data.table(summary(modTfullrem0)$tTable)
coefs5 <- as.data.table(summary(modTfullJbetarem0)$tTable)
coefs6 <- as.data.table(summary(modTfullHornrem0)$tTable)
coefs$mod <- 'Jtu'
coefs2$mod <- 'Jbeta'
coefs3$mod <- 'Horn'
coefs4$mod <- 'Jtu rem0'
coefs5$mod <- 'Jbeta rem0'
coefs6$mod <- 'Horn rem0'
coefs$var <- rownames(summary(modTfull1)$tTable)
coefs2$var <- rownames(summary(modTfullJbeta)$tTable)
coefs3$var <- rownames(summary(modTfullHorn)$tTable)
coefs4$var <- rownames(summary(modTfullrem0)$tTable)
coefs5$var <- rownames(summary(modTfullJbetarem0)$tTable)
coefs6$var <- rownames(summary(modTfullHornrem0)$tTable)
# extract temperature effects and bind model coefs together
cols <- c('var', 'Value', 'Std.Error', 'mod')
allcoefsfull <- rbind(coefs[grep('temptrend|REALM', var), ..cols],
coefs2[grep('temptrend|REALM', var), ..cols],
coefs3[grep('temptrend|REALM', var), ..cols],
coefs4[grep('temptrend|REALM', var), ..cols],
coefs5[grep('temptrend|REALM', var), ..cols],
coefs6[grep('temptrend|REALM', var), ..cols])
allcoefsfull$var[allcoefsfull$var == 'temptrend_abs.sc'] <- 'temptrend_abs.sc:REALMFreshwater'
# add average temperature effect (across realms) to realm-specific temperature effects
meantempeffect <- allcoefsfull[var == 'temptrend_abs.sc:REALMFreshwater', mean(Value), by = mod]$V1
allcoefsfull[var == 'temptrend_abs.sc:REALMMarine', Value := Value + meantempeffect]
allcoefsfull[var == 'temptrend_abs.sc:REALMTerrestrial', Value := Value + meantempeffect]
# remove non-temperature effects
allcoefsfull <- allcoefsfull[grepl(':', allcoefsfull$var) & grepl('temptrend', allcoefsfull$var), ]
# add info for plotting
allcoefsfull$lCI <- allcoefsfull$Value - allcoefsfull$Std.Error # lower confidence interval
allcoefsfull$uCI <- allcoefsfull$Value + allcoefsfull$Std.Error
nvar <- nrow(allcoefsfull)/3
allcoefsfull$y <- 1:nvar + rep(c(0, 0.1, 0.2), c(nvar, nvar, nvar)) # y-values
# clean up some variable names for nicer plotting
allcoefsfull$varname <- gsub('temptrend_abs.sc:|temptrend.sc:', '', allcoefsfull$var)
allcoefsfull$varname <- gsub('REALM|REALM2', '', allcoefsfull$varname)
allcoefsfull$varname <- gsub('.sc', '', allcoefsfull$varname)
allcoefsfull$varname <- gsub('_bowler', '', allcoefsfull$varname)
allcoefsfull$varname <- gsub('tsign1', 'warming', allcoefsfull$varname)
allcoefsfull$varname <- gsub('tsign-1', 'cooling', allcoefsfull$varname)
# set x-axes
xlims1 <- c(-0.01, 0.11) # for realms
xlims2 <- c(-0.05, 0.05) # for traits
xlims3 <- c(-0.01, 0.02) # for environment
xlims4 <- c(-0.016, 0.01) # for community
xlims5 <- c(-0.006, 0.005) # for human
ddg <- 0.5 # vertical dodge for each model
# choose which variables are in which graph
set1 <- c('Terrestrial', 'Marine', 'Freshwater')
set2 <- c('mass', 'speed', 'lifespan', 'consumerfrac', 'endothermfrac', 'tempave_metab')
set3 <- c('seas', 'microclim', 'tempave')
set4 <- c('npp', 'nspp', 'tsign-1:thermal_bias', 'tsign1:thermal_bias')
set5 <- c('human:TerrFresh', 'human:Marine')
p1 <- ggplot(subset(allcoefsfull, varname %in% set1),
aes(varname, Value, group = mod, color = mod)) +
geom_hline(yintercept = 0, linetype = 'dashed', color = 'light grey') +
geom_errorbar(aes(ymin = lCI, ymax = uCI), width = 0, position = position_dodge(ddg)) +
geom_point(position = position_dodge(ddg)) +
labs(y = 'Temperature change effect', x = '', tag = 'A') +
scale_color_brewer(palette = 'Dark2') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='none',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_flip(ylim = xlims1)
p2 <- ggplot(subset(allcoefsfull, varname %in% set2),
aes(varname, Value, group = mod, color = mod)) +
geom_hline(yintercept = 0, linetype = 'dashed', color = 'light grey') +
geom_errorbar(aes(ymin = lCI, ymax = uCI), width = 0, position = position_dodge(ddg)) +
geom_point(position = position_dodge(ddg)) +
labs(y = 'Interaction with temperature change effect', x = '', tag = 'B') +
scale_color_brewer(palette = 'Dark2') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='none',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_flip(ylim = xlims2)
p3 <- ggplot(subset(allcoefsfull, varname %in% set3),
aes(varname, Value, group = mod, color = mod)) +
geom_hline(yintercept = 0, linetype = 'dashed', color = 'light grey') +
geom_errorbar(aes(ymin = lCI, ymax = uCI), width = 0, position = position_dodge(ddg)) +
geom_point(position = position_dodge(ddg)) +
labs(y = 'Interaction with temperature change effect', x = '', tag = 'C') +
scale_color_brewer(palette = 'Dark2') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='none',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_flip(ylim = xlims3)
p4 <- ggplot(subset(allcoefsfull, varname %in% set4),
aes(varname, Value, group = mod, color = mod)) +
geom_hline(yintercept = 0, linetype = 'dashed', color = 'light grey') +
geom_errorbar(aes(ymin = lCI, ymax = uCI), width = 0, position = position_dodge(ddg)) +
geom_point(position = position_dodge(ddg)) +
labs(y = 'Interaction with temperature change effect', x = '', tag = 'D') +
scale_color_brewer(palette = 'Dark2') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='none',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_flip(ylim = xlims4)
p5 <- ggplot(subset(allcoefsfull, varname %in% set5),
aes(varname, Value, group = mod, color = mod)) +
geom_hline(yintercept = 0, linetype = 'dashed', color = 'light grey') +
geom_errorbar(aes(ymin = lCI, ymax = uCI), width = 0, position = position_dodge(ddg)) +
geom_point(position = position_dodge(ddg)) +
labs(y = 'Interaction with temperature change effect', x = '', tag = 'E') +
scale_color_brewer(palette = 'Dark2') +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
panel.background = element_blank(), axis.line = element_line(colour = "black"),
legend.position='bottom',
axis.text=element_text(size=7),
axis.title=element_text(size=7)) +
coord_flip(ylim = xlims5)
grid.arrange(p1, p2, p3, p4, p5, ncol = 2, layout_matrix = rbind(c(1,2), c(3,4), c(5, NA)))
NA
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